Sicheng Shen

BI & Data Product Manager · Governed Data Platforms · Production AI-Data Agents · US Last-Mile Marketplace + Cross-Border e-Commerce

Wood-Ridge, NJ · sichengshenpersonal@gmail.com · linkedin.com/in/sichengshen

Summary

Product manager who architected and owns Speed Xpress's end-to-end governed data platform serving 280+ active users, 80K+ daily API calls, and all C-level stakeholders at a 400-person US last-mile logistics marketplace processing 800K+ daily labels (peak 1.6M+) for cross-border Chinese platforms + US domestic customers across own-fleet, PUDO, and 3rd-party partnership (Maersk, USPS) modes. Leads 8-person BI + data engineering team on 2–3 releases per sprint cadence. Direct 1st-level data ownership across all HQ functional departments. Signature: turning chaotic operational data into trustworthy, auditable, scalable systems.

Experience

BI & Data Product Manager — Speed Xpress Inc.

Jun 2025 – Present

Owns product strategy, roadmap, and end-to-end delivery of SpeedX's internal data platform (Data Portal, Executive & Functional Dashboards, AI Data Agent, Data Asset Management, Access Control, Feedback System). Manages 8-person BI + data engineering team (BI Head/Scrum Master, 2 Data Analysts, 1 Architect, 2 Data Engineers, 1 DevOps, 1 QA; US regional team expanding).

Data Portal (from-scratch architecture): Designed and shipped unified internal data platform now serving 280+ active users with 80,000+ daily API calls, 300+ daily core-report clicks, and 5-minute average daily user time — replacing prior manual reconciliation across marketplace SLA, driver performance, and multi-region billing.
SPX Data Agent (production-grade AI-data agent): Led dual-routing architecture (Spec-Driven deterministic engine + Free-Agent fallback via GPT-4o) with 6-layer context system, 163+ data tables, 95 business terms / 343 aliases, 46 pre-validated SQL templates, 13-field PII masking, and multi-layer SQL guardrails; 15 Spec contracts across 5 domains, full request-level observability.
Company-Wide KPI Governance: Standardized 20+ operational metrics across 6 functional lines — QA (INR, DNR, long-tail parcels), Warehouse Ops (driver same-day delivery, warehouse throughput), Customer Service (exception packages), Middle Mile (dispatch on-time rate, Dispatch 2.0 adoption), Finance (selling rate, claims), Commercial (SLA compliance) — driving cross-functional alignment across 14 departments as single source of truth for the Executive Dashboard.
Marketplace SLA + OTP Framework (multi-mode): Translated contractual SLA and operational performance standards for own-fleet last-mile + PUDO + 3rd-party partnerships (Maersk, USPS) into SQL-based metric definitions, threshold classifications, and automated scheduled reports delivered to ops and commercial stakeholders.
ADP Access Control & 3-Layer Governed Query Framework: Designed team-based permission architecture, query-level auditability, and 3-layer analytical access model (SQL Builder for controlled deep investigation / Data Builder for guided exploration / Pre-defined Data Requests for standardized queries) — separating management-layer approval from technical execution.

Support Manager — Speed Xpress Inc.

Prior tenure

Led U.S. technical support operations for proprietary internal systems, owning daily task planning, SOPs, escalation, and complex cross-domain case coordination. Served as primary coordination point across Support, Operations, Product, and Engineering. Long-term involvement in support-side database investigations seeded the current data-governance and access-control agenda now owned in PM capacity.

Education

University of Pennsylvania · Master of Science in Engineering, Systems Engineering

Sep 2023 – May 2025

Oberlin College · B.S. Mathematics (Honors) · B.S. Economics · Double Minors in CS & Statistical Modeling

Sep 2019 – May 2023

Technical & Product Stack

SQL · Python (Flask, Jinja2, python-docx) · MySQL · Apache Pinot · YAML · Figma · Notion · GPT-4o (production integration) · LLM guardrail design · PII/PHI-adjacent governance patterns · Agile (2–3 releases per sprint)

Sicheng Shen

BI & Data Product Manager · Governed Data Platforms · Production AI-Data Agents · US Last-Mile Marketplace + Cross-Border e-Commerce

Wood-Ridge, NJ · sichengshenpersonal@gmail.com · linkedin.com/in/sichengshen

Summary

Product manager who architected and owns Speed Xpress's end-to-end governed data platform serving 280+ active users, 80K+ daily API calls, and all C-level stakeholders at a 400-person US last-mile logistics marketplace processing 800K+ daily labels (peak 1.6M+) for cross-border Chinese platforms + US domestic customers across own-fleet, PUDO, and 3rd-party partnership modes (Maersk, USPS). Leads 8-person BI + data engineering team on 2–3 releases per sprint cadence. Direct 1st-level data ownership across all HQ functional departments. Signature: turning chaotic operational data into trustworthy, auditable, scalable systems — with a growing production practice in AI-data governance.

Experience

BI & Data Product Manager — Speed Xpress Inc.

Jun 2025 – Present

Owns product strategy, roadmap, and end-to-end delivery of SpeedX's internal data platform — spanning the Data Portal, AI Data Agent, Executive & Functional Dashboards, Autonomous Data Platform (ADP), Data Asset Management (DAM), Access Control, and Feedback System. Manages 8-person BI + data engineering team (1 BI Head/Scrum Master, 2 Data Analysts, 1 Architect, 2 Data Engineers, 1 DevOps, 1 QA); US regional team expanding with additional QA, DevOps, and Data Engineer. Serves as direct data-partner to every HQ functional department (Ops, Finance, Commercial, Engineering, Customer Service).

Data Portal — Product Design, File System Architecture & 4-Phase Roadmap: Designed and led the complete product architecture and file system of the Data Portal from scratch — defining module scope, content ownership, data relationships, and governance boundaries across Executive Dashboards, Functional Team Dashboards, Data Asset, AI Data Agent, Access Control, and Feedback System. Authored a four-phase delivery roadmap sequencing platform delivery against organizational readiness. Live impact: 280+ active users, 80,000+ daily API calls, 300+ daily core-report clicks, 5-minute average daily user time.
SPX Data Agent — AI-Powered Logistics Analytics Product: Defined a dual-routing architecture: (a) a Spec-Driven deterministic path constraining the LLM to parameter extraction + narrative synthesis while a Python engine handles all numerical computation, and (b) a Free-Agent path (GPT-4o, up to 10 reasoning iterations, 20 tools) for long-tail investigations. Designed a 6-layer context system (Schema Metadata, Semantic Layer, SQL Templates, Spec Contracts, Learning Memory, Runtime Context) covering 163+ data tables, 95 business terms / 343 aliases, and 46 pre-validated SQL templates across MySQL BI and Apache Pinot. Oversaw 15 Spec contracts across 5 business domains (Delivery, Aging, Billing, Driver Performance, Operations) enabling YAML-based code-free expansion. Defined multi-layer SQL guardrails, 13-field PII masking applied pre-return, full request-level observability, and step-guided feedback + self-correction loops.
Autonomous Data Platform (ADP) — Access Control & Data Product Pattern Design: Designed the team-based permission architecture governing all Data Portal modules, establishing role-scoped access, query-level auditability, and deliberate separation between management-layer approval (human-governed) and technical execution (system-enforced). Defined standardized interfaces for metric definitions, report templates, dashboard layouts, and analytical output formats — ensuring consistent UX and reuse across Executive Dashboard, Functional Team Dashboards, and AI Agent.
Executive Dashboard & Company-Wide KPI Governance: Standardized 20+ operational metrics across 6 functional lines: QA (INR / lost & stolen, DNR, long-tail parcels, physical exceptions), Warehouse Ops (driver same-day delivery rate, warehouse throughput), Customer Service (exception packages), Middle Mile (dispatch on-time, Dispatch 2.0 adoption), Finance (selling rate, claims), Commercial (SLA compliance). Drove structured cross-functional agreement across 14 departments as single source of truth. Established a reusable governance baseline enabling all downstream BI product development.
3-Layer Governed Query Framework: Designed calibrated data access — SQL Builder (controlled deep investigations, cross-database extraction by qualified users), Data Builder (guided self-service exploration), Pre-defined Data Requests (standardized high-frequency needs with pre-approved logic). Defined permission configuration, access logging, and query tracking protocols to establish traceability foundation for future compliance and audit review.
Marketplace SLA + OTP Reporting Framework (multi-mode): Translated contractual SLA and operational performance standards for own-fleet last-mile + PUDO + 3rd-party partnerships (Maersk, USPS) into SQL-based metric definitions, threshold classifications, and automated scheduled reports for operations and commercial stakeholders across cross-border and US domestic corridors.
Python Reporting Stack: Built end-to-end automated report generation with Flask + Jinja2 + python-docx — scheduled query execution against live databases, dynamic template rendering, formatted Word/HTML report assembly. Authored modular reporting libraries for scheduled execution, result transformation, template rendering, and document output across ops, billing, and multi-region logistics.
Internal AI Knowledge Base: Coordinated with Operations, Finance, Engineering, Legal, Commercial, Customer Service to consolidate 55+ non-confidential documents across 14 departments as inputs to an AI-powered knowledge assistant; owns ongoing testing, content quality, and updates.
Enterprise Ticketing System Evaluation & Rollout: Led org-wide deployment across HR, Billing, Procurement, Tech Support, DevOps, and IT Support — embedding ticket-based workflows into daily operations and improving efficiency, transparency, and analytical readiness.

Support Manager — Speed Xpress Inc.

Prior tenure

Led U.S. technical support operations for proprietary internal systems, owning daily task planning, priority management, SOPs, escalation workflows, and coordination of complex cross-domain cases. Served as primary coordination point between Support and Operations, Product, and Engineering. Drove evaluation, rollout, and adoption of support-facing internal platforms (ticketing, knowledge, support portals).

Contributed to company-wide analytical data governance and internal data-access standards from sustained support-side database investigation experience — permission models, access traceability, KPI definitions, usage guidelines — seeding the current data platform agenda now owned in PM capacity.

Education

University of Pennsylvania — Master of Science in Engineering, Systems Engineering

Sep 2023 – May 2025

Oberlin College — B.S. in Mathematics (Honors) · B.S. in Economics · Double Minors in Computer Science & Statistical Modeling

Sep 2019 – May 2023

Technical & Product Stack

SQL · Python (Flask, Jinja2, python-docx, modular libraries) · MySQL · Apache Pinot · YAML · Figma · Notion · GPT-4o (production integration) · LLM guardrail design · PII/PHI-adjacent governance patterns · Agile (2–3 releases/sprint) · Cross-border logistics operations · Marketplace SLA/OTP frameworks · Multi-region billing systems

Tailored Bullet Library

Substitute these variants into the master resume when submitting to specific Amazon orgs

用法：在 base resume 里，把 Summary 换成 org-specific 版本；bullets 里前 2-3 条按 org 变种替换或排序。保持数字一致。

Priority 1 · Amazon Logistics (AMZL / ALT — Last Mile)

Summary (tailored)

Product manager who architected and owns Speed Xpress's end-to-end US last-mile marketplace data platform — SLA framework, driver performance scorecards, OTP tracking, and multi-region billing — across own-fleet + PUDO + 3rd-party partnership modes (Maersk, USPS) at 800K+ daily labels (peak 1.6M+). Directly parallels the DSP marketplace and delivery-station tech problem shape at Amazon Logistics, one order of magnitude smaller. Leads 8-person BI + data engineering team; direct data ownership across all HQ functional departments.

Bullet swaps (use in top 3 position)

Move these to the top of Experience bullets:

Last-Mile Marketplace SLA + Driver Performance Framework: Designed the SLA / OTP / driver-performance scorecard system for a two-sided last-mile fleet marketplace serving cross-border and US domestic corridors — same problem shape as Amazon DSP scorecard (on-time / acceptance / disruption / app-usage), executed at 800K+ daily labels scale across own-fleet, PUDO, and 3rd-party partnerships (Maersk, USPS). Standardized 20+ metrics across QA (INR, DNR, long-tail), Warehouse Ops (driver same-day delivery rate), Middle Mile (dispatch on-time rate), and Ops Center (Dispatch 2.0 adoption).
Multi-Region Billing & Unit Economics: Built cross-mode settlement + billing product covering own-fleet driver payments, PUDO cost accounting, and 3rd-party partner (Maersk/USPS) rate reconciliation. Direct parallel to DSP settlement, Flex payment, and station-level cost accounting at AMZL.
Executive Dashboard + Station-Level Analytics: Ships live to all C-level (300+ daily clicks, 5-min avg daily user time, 280+ active users). Mirrors delivery-station manager dashboards + AMZL executive rollup surface.

1-line pitch

"I built US last-mile marketplace SLA + OTP + driver performance + multi-region billing at 1/50th of AMZL scale. Your DSP marketplace + delivery station tech is my domain scaled up 50×."

Priority 2 · AWS QuickSight / Quick Suite (Title Match)

Summary (tailored)

BI & Data Product Manager who architected and owns Speed Xpress's internal analog to QuickSight — a self-serve data platform serving 280+ active analyst + operator users at 80K+ daily API calls, with an embedded Executive Dashboard, a governed 3-layer query framework, and a production Amazon-Q-analog AI agent (dual-routing, 6-layer context, 46 SQL templates, 13-field PII masking). Has lived every objection, adoption blocker, and success metric a QuickSight buyer faces. Signature: metric drift, semantic-layer governance, and permissions-aware NL Q&A on executive dashboards.

Bullet swaps

Internal QuickSight Analog — Data Portal (from-scratch): Architected and shipped self-serve BI + analytics platform serving 280+ active users, 80,000+ daily API calls, 300+ daily core-report clicks, 5-minute average daily user time — including embedded Executive Dashboard (equivalent to Q Executive Summaries), Business/Ops dashboards, Data Asset catalog (Quick Index analog), and 3-layer governed query framework (Topics + Quick Suite governance analog).
Amazon Q Analog — Production AI Data Agent: Designed and led production-grade NL Q&A over warehouse data with dual-routing (deterministic Spec-Driven + Free-Agent GPT-4o), 6-layer context (Schema, Semantic, SQL Templates, Spec Contracts, Learning Memory, Runtime), 46 pre-validated SQL templates, 13-field PII masking, and step-guided user feedback + agent self-correction — directly analogous to the grounding, hallucination, and permissions-aware trust problem the Q in QuickSight team is solving.
Company-Wide KPI Standardization (metric drift at source): Standardized 20+ operational metrics across 14 departments — the #1 unsolved problem in enterprise BI. Direct match for Structured Data Analytics PMT and Quick Identity Governance product surfaces.

1-line pitch

"I built the internal version of what QuickSight ships externally — I'm the mid-market customer QuickSight is trying to win. Watching the Quick Suite launch felt like watching my roadmap in a bigger room."

Priority 3 · Amazon Grocery (Same-Day Delivery + WWGST)

Summary (tailored)

Same-day marketplace ops product manager who architected and owns Speed Xpress's end-to-end SLA / OTP / driver performance / multi-region billing infrastructure — the exact problem shape Amazon Same-Day Grocery Delivery is scaling 10× with cold-chain layered on top. Processes 800K+ daily labels (peak 1.6M+) across own-fleet last-mile + PUDO + 3rd-party partnership modes (Maersk, USPS). Leads 8-person BI + data engineering team.

Bullet swaps

Same-Day Slot Compliance + OTP Framework: Designed marketplace SLA framework and slot-compliance tracking for a same-day-adjacent US last-mile marketplace — directly analogous to Amazon Grocery's same-day slot SLAs and Whole Foods 2-hour window OTP. Standardized 20+ operational metrics (driver same-day delivery, dispatch on-time, warehouse throughput) across 14 departments.
Multi-Region Grocery-Adjacent Unit Economics: Built multi-region billing + cost accounting product for cross-mode operations (own-fleet driver payments, PUDO cost, 3rd-party partner rate reconciliation) — same shape as same-day grocery unit economics (Prime Free Delivery $9.99 subscription economics). Grocery delivery margins are razor-thin; I've built the data platform that makes those decisions defensible.
Unified Data Spine (WWGST Supplier Performance analog): Executive Dashboard + KPI Governance across all HQ functional departments (Ops, Finance, Commercial, CS) directly parallels the "One Grocery" cross-Fresh/WFM/Amazon.com KPI standardization problem now under Udit Madan's ops org.

1-line pitch

"Amazon's Same-Day Grocery is my SPX problem at 10× scale with cold-chain layered on top — same-day slot SLAs, driver marketplace unit economics, unified private-label spine. That's the room I want to be in."

Priority 4 · Amazon Global Logistics (AGL)

Summary (tailored)

Cross-border logistics data PM who architected and owns Speed Xpress's end-to-end marketplace SLA / OTP / multi-region billing platform for a US last-mile marketplace serving Chinese cross-border e-commerce platforms + US domestic customers. Bilingual (English + Chinese), China-born, with direct production experience across the exact China → US corridor AGL now doubling down on (Vietnam Nov 2025, GWD Shenzhen Mar 2026). Ships 800K+ daily labels (peak 1.6M+).

Bullet swaps

Cross-Border Marketplace SLA + Multi-Region Billing: Built the SLA framework + multi-region billing product for cross-border corridors (Chinese origin platforms → US last-mile) across own-fleet, PUDO, and 3rd-party partnership modes (Maersk, USPS) — direct pattern match for AGL Pricing (bundled per-shipment quotes, FBA fee interplay) and Shipper Experience (tracking + visibility surface in Seller Central).
Governed AI Over Cross-Border Ops Data: Production-grade AI-data agent with 6-layer context, 46 SQL templates, 13-field PII masking, multi-layer SQL guardrails — the same governance + safety problem AGL is solving with PreDepart AI customs classification (CBP CSOP-certified). Rare skill for a PM candidate.
Bilingual + China-Born Domain Fluency: Direct language + cultural context for AGL's #1 origin (China) and #2 origin (Vietnam). Lived inside the Chinese-seller-to-US delivery experience Amazon is doubling down on via GWD Shenzhen and Fulfill From Origin.

1-line pitch

"I've built the exact KPI governance + data platform + AI safety layer AGL needs, one order of magnitude smaller, on the exact China-to-US corridor AGL is doubling down on."

Priority 5 · Amazon Ads (AMC · Sponsored · DSP)

Summary (tailored)

Data-platform product manager who standardized KPIs and governance across a two-sided marketplace with 800K+ daily transactions — the closest operational analog to a data clean-room + auction environment. Built and owns Speed Xpress's marketplace SLA framework, pricing / billing surface, and governance layer (13-field PII masking, query audit trails, 3-layer permission architecture) — direct architecture match for AMC clean-room design and advertising measurement infrastructure. Ready to apply the same rigor to Amazon Ads revenue-side surfaces.

Bullet swaps

Marketplace Auction Dynamics (SPX SLA + Pricing): Owned marketplace SLA + pricing / billing framework across a two-sided fleet marketplace with quality signals, throttling, and reserve pricing — directly translatable to Sponsored Products auction dynamics, DSP unified Campaign Manager, and bid-mechanism design. 20+ standardized metrics across 14 departments.
Clean-Room-Grade Data Governance: Built pseudonymization, permission architecture, audit-trail, and cross-region access controls for a data platform serving 280+ users across all HQ functional departments — same architectural stack AMC requires for privacy-safe advertiser data collaboration. 13-field PII masking applied pre-return with full request-level observability.
KPI Standardization (ROAS-drift analog): Drove cross-departmental agreement on 20+ metric definitions — the exact problem shape as ROAS vs. iROAS vs. NTB-ROAS vs. incrementality standardization inside Amazon Ads across DSP + Sponsored + AMC. PMs who can impose KPI consistency at scale are rare.

1-line pitch

"Data-platform PM who standardized KPIs + governance across a marketplace auction. Ready to apply the same rigor to AMC clean-room design + Ads measurement + revenue-side auction products."

Priority 6 · Amazon Health Services (Pharmacy · One Medical)

Summary (tailored)

Governed-data product manager who architected and owns the exact regulated-data pattern stack Amazon Health requires — 13-field PII masking applied pre-return, query audit trails, team-based permission architecture, and a 3-layer governed query framework — at a US last-mile logistics marketplace. Uniquely combines HIPAA/DEA-adjacent governance rigor with last-mile delivery domain expertise (Rx same-day Rx to 4,500 cities is same-day marketplace ops with regulated overlay). Ships 800K+ daily transactions across own-fleet + PUDO + 3rd-party partnerships.

Bullet swaps

Regulated-Data Governance Stack (HIPAA/DEA-adjacent): Architected 13-field PII masking (pre-return enforcement), full query audit trails, RBAC across 6 functional lines, and 3-layer governed query framework — the exact pattern HIPAA §164.312(b) audit controls and DEA CSA record-keeping require. Same architectural stack that would have prevented PillPack's 2024 $300K CSA settlement.
Last-Mile Marketplace Ops at Scale: Built SLA / OTP / driver performance / multi-region billing for 800K+ daily labels across own-fleet + PUDO + 3rd-party partnerships (Maersk, USPS) — directly translatable to Amazon Pharmacy same-day Rx delivery (4,500 cities + drone) unit economics.
Governed AI over Operational Data: Production dual-routing AI-data agent with 6-layer context, 46 SQL templates, multi-layer guardrails, and step-guided self-correction — same architectural pattern as Amazon Pharmacy's 90%-faster GenAI-driven Rx processing (Bedrock + Textract + SageMaker) and RxPass adherence workflows.

1-line pitch

"I own governed data for a last-mile logistics marketplace AND I speak HIPAA. Regulated-data governance + last-mile Rx delivery — the exact intersection Amazon Health is scaling. That's my signature strength at maximum leverage."