AC Ashwin Chembu

Bay Area · Open to building what matters

I build thoughtful software where product, AI, and data meet.

I build practical AI-enabled products, backend systems, and thoughtful user experiences—from document intelligence and automation to real-time, data-driven applications.

Showing a balanced view of secure integrations, retrieval-grounded products, data pipelines, and clearly labeled supporting work.

See selected work ↓ View résumé ↗
React✦Next.js✦TypeScript✦Python✦FastAPI✦Node.js✦Express✦REST APIs✦OpenAPI✦MCP✦WebSockets✦RAG✦LangChain✦AWS Bedrock✦Pinecone✦FAISS✦SQL✦Snowflake✦MongoDB✦Firebase✦AWS✦Three.js✦Playwright✦

01 / Featured case studies

Practical systems, built end to end.

Three representative case studies spanning secure handoffs, retrieval-grounded product work, and multi-tenant SaaS delivery.

Technology map

A compact view of the stack.

InterfacesReact · TypeScript
ServicesPython · FastAPI · Node.js · Express
Data & cloudMongoDB · Firebase · AWS · S3
Applied AIRAG · Vector search · Grounding
IntegrationsREST APIs · OpenAPI · MCP · Webhooks
Quality & deliverypytest · Vitest · Playwright · Render

Supporting work

Prototypes and collaborations, labeled clearly.

Collaborative research/classroom work

158 attributed commits · 6 public resources · 12K+ aggregate uses

Tech4Good FigJam Toolkit

Built and iterated chat, text-entry, polling, carousel, radar, Likert, tooltip, Excel/XLSX, and shared Node/MongoDB workflows; analyzed 43K+ anonymized interaction records with Python/Pandas.

TypeScript · Figma API · Node.js/Express · MongoDB · Excel/XLSX · Python/Pandas
Chat ↗Text Entry ↗Polling ↗GIF ↗Quote ↗Wordcloud ↗Shared source ↗All resources ↗

The six Figma listings and 12K+ aggregate usage are team-published outcomes; GitHub attributes 158 commits to me in the collaborator-owned public repository.

Private organization product

Full Android app · Multi-tenant SaaS

Sylk

Built and packaged the complete supply-chain product as a React/Capacitor Android application spanning onboarding, authentication, orders, inventory, billing, chat, files, team access, and phone-specific navigation.

React 19 · TypeScript · Firebase · Capacitor · Android · Playwright

Source and customer data remain private; Google Play publication is not claimed.

Private system · Generalized evidence

Agentic automation · Human approval

Human-in-the-Loop Job Operations

Maps job requirements to verified evidence, generates reproducible résumé specifications, checks ATS and rendered PDFs, and requires exact-document approval before submission.

Node.js · Python · JSONL · Excel · Telegram · PDF QA

Application records, email data, credentials, and job-search artifacts are never published.

Anonymized employer evidence

Healthcare data · MDM

Healthcare Data Quality & MDM

Built Snowflake and layered ETL/MDM workflows across 327K+ records, including entity matching across 67K+ records, reconciliation, quality gates, and stewardship review.

Snowflake · SQL · ETL / MDM · Entity resolution · Data quality

No client identity, private data, SQL, screenshots, or proprietary artifacts are published.

Anonymized employer evidence

Salesforce · CRM delivery

CRM Workflow & UAT Delivery

Turned stakeholder feedback, CRM metadata, and live behavior into requirements, UAT evidence, acceptance criteria, backlogs, user guides, and implementation-ready handoffs.

CRM metadata · UAT · React/Vite · Decision logs · User guides

Synthetic abstraction only; no client identity, screenshots, transcripts, or credentials are published.

Figma-originated UI prototype

Frontend · Three.js

3D Commerce Experience

Implemented a responsive varsity-jacket storefront and GLB/Draco product configurator with independent material, color, and placement controls.

React · TypeScript · Three.js · Vite
Open preview ↗Source ↗
Figma-originated UI prototype

Frontend · Healthcare workflows

Enterprise Healthcare Dashboard

Translated account hierarchies, trial progress, follow-ups, barriers, and task-editing workflows into reusable React/TypeScript and Vue interface patterns.

React · TypeScript · Vue · Tailwind
View source ↗
Founder product + HackDavis collaboration

Education AI · Clearly attributed

Tallyrus + AbleGrade

Tallyrus applies RAG and educator-defined rubrics to essay feedback; AbleGrade was a separate co-built HackDavis 2024 prototype for ESL writing support, PDF parsing, and progress tracking.

Python · LangChain · Pinecone · MERN · GPT-4
Visit Tallyrus ↗

Both repositories are private; no source link is published.

Public repository

Applied AI · Document search

Azure RAG Document Search

Built multi-format ingestion, chunking, Azure OpenAI embeddings, FAISS retrieval, source citation, and interactive Q&A across a 30+ document corpus.

Python · Azure OpenAI · LangChain · FAISS
View source ↗
Public repository · Inspect before reuse

Voice AI · AWS data workflows

Nova Sonic Assistant

Built real-time voice sessions, browser and iPad clients, transcript and tool-log interfaces, plus S3, Glue, Redshift, and call-analysis workflows.

Node.js · Vue · WebSockets · Bedrock · S3 · Glue
View source ↗
Collaborative research work

UC Davis · Digital verification

Echo

Collaborated on identity-authentication and secure-media workflows using React, Node, Flask, MongoDB, AWS S3, and a research blockchain stack.

React · Flask · MongoDB · AWS S3
Read article ↗View GitHub ↗Watch demo ↗

Earlier work

Systems, coursework, and experiments.

Coursework · CUC Davis Operating SystemsShells, threads, semaphores, and a FAT-style file system ↗ Coursework · Programming languagesECS 140 Project SetJava, Haskell, Prolog, and Erlang ↗ Take-home project · C++Six Degrees ApplicationCrow backend and React/TypeScript interface ↗ Early web experimentRetro ArcadeBrowser game loops, state, collision, and scoring ↗ Frontend prototype · MapsRideshare Map InterfaceReact routing and Google Maps components ↗ Fundamentals archiveAlgorithms & Data StructuresC++, C, Python, assembly, hash tables, and systems exercises ↗

02 / Experience

Engineering across the stack.

Building at the intersection of technology, operations, and user needs.

Jun 2024—Present
ZS logo

ZS Associates

Business Technology Solutions Associate

Cross-functional scope across AI/RAG services, Snowflake and MDM, API integrations, CRM validation, and delivery documentation.

  • Built enterprise AI/RAG assistants and backend services with React, FastAPI, AWS data services, Redis, and document-ingestion workflows.
  • Developed Snowflake SQL pipelines for master-data extraction, identifier matching, reconciliation, and quality reporting across 327K+ records.
More scope & impact
  • Designed entity-matching and license-reconciliation frameworks across 67K+ records using healthcare identifiers and fuzzy-search signals for stewardship review.
  • Defined layered ETL/MDM flows with landing, staging, transformation, canonical, and publication layers, including delta logic, history, and data-quality controls.
  • Built secure browser-to-backend workflows for targeted entity lookup and asynchronous CSV/XLSX enrichment while keeping credentials server-side and write operations explicitly gated.
  • Implemented and validated React/Vite business workflows for hierarchical records, approval states, read-only gating, linked actions, and executive summaries.
  • Translated ambiguous stakeholder requirements into testable rules, API integrations, source-to-target mappings, UAT scenarios, decision logs, executive decks, and delivery documentation.
  • Built Python and Flask endpoints plus AWS S3 and Smartsheet API integrations for automated document-management workflows.
  • Contributed to RAG evaluation and model workflows using FastAPI, Uvicorn, pytest, and enterprise AI tooling.
PythonFastAPISnowflakeAWSETL / MDMCRM / UATApplied AI
Jun 2023 — Now
Tallyrus logo

Tallyrus

Founder & Lead Developer

  • Built NLP and retrieval-augmented essay feedback with Python, LangChain, Pinecone, OpenAI, and educator-defined rubrics.
  • Developed Flask, MongoDB, and S3 services alongside React grading workflows and iterative educator feedback loops.
More product & founder details
  • Built role-aware classroom, assignment, submission, and rubric workflows with student/teacher authorization before changing academic records or files.
  • Implemented React/TypeScript and Express workflows with Google OAuth, protected routes, secure sessions, CORS, rate limiting, and security headers.
  • Added S3-backed upload, replacement, and document-parsing workflows for PDFs, DOCX files, rubrics, and prior graded essays.
  • Led product planning, user testing, iterative development, and fundraising, raising five figures for product development.
RAGLangChainPineconeFlaskMongoDBOAuthS3
Visit Tallyrus ↗
Mar 2023 — Now
Tech4Good Lab logo

Tech4Good Lab

Data Analyst / Web Development Lead

  • Contributed 158 attributed GitHub commits across a collaborator-owned FigJam widget suite for research and classroom workflows.
  • Built and iterated chat, text-entry, polling, carousel, radar, Likert, tooltip, and Excel/XLSX modules; the six team-published Community resources show 12K+ aggregate uses.
More scope, evidence & links
  • Implemented threaded and synchronized chat, edits, anonymous responses, bulk loading, voting, multi-select polls, vote changes, grouped widgets, and configurable admin controls.
  • Connected Node.js, Express, and MongoDB persistence and logging routes with Python/Pandas analysis over 43K+ anonymized interaction records.
  • Contributed a responsive Angular top-paying-industries component for Tech4Good’s career-exploration product.
Figma APITypeScriptNode.jsExpressMongoDBExcel/XLSXPython/PandasAngular
Chat ↗Text Entry ↗Polling ↗GIF ↗Quote ↗Wordcloud ↗Shared source ↗Team profile ↗
Sep 2023 — Mar 2024
UC Davis logo

Echo · UC Davis

Software Engineer & Research Collaborator

  • Collaborated with Professor Mohammad Sadoghi’s research group on digital identity authentication and secure-media workflows.
  • Integrated facial-detection APIs, AWS S3/Multer storage, MongoDB, Express, Flask, and Python SDK services.
More engineering details
  • Built React and Node interfaces around capture, matching, proof creation, and verification states.
  • Connected the end-to-end workflow to ResilientDB for identity-proof creation and persistence.
ReactFlaskAWS S3ResilientDB
Read project article ↗View GitHub ↗Watch demo ↗
Open full résumé ↗ Verify on LinkedIn ↗

03 / About

Curious by default.
Practical by design.

I’m a UC Davis computer science graduate based in Fremont, California. I like untangling hard technical problems, but I care just as much about whether the result feels intuitive and earns its place in someone’s workflow.

My work spans product engineering, grounded AI applications, secure backend integrations, and data systems. The common thread is simple: make complex technology useful, testable, and clear.

University of California, DavisB.S. Computer Science
GraduatedJune 2025 · GPA 3.7
CredentialsAWS Cloud Practitioner · Google Data Analytics
LeadershipTau Beta Pi Treasurer · Dean’s Honor

Core capabilities

01

Product engineering

React, Next.js, TypeScript, Node.js, responsive interfaces, end-to-end product delivery

02

Applied AI

RAG, vector search, evaluation pipelines, LangChain, OpenAI, AWS Bedrock

03

Backend & data

Python, FastAPI, Flask, SQL, MongoDB, Snowflake, ETL, AWS

04

Integration & quality

REST APIs, OpenAPI, MCP, webhooks, authentication, Playwright, Vitest, pytest

05

Automation & delivery

Human-in-the-loop workflows, audit trails, ATS and PDF quality gates, UAT, decision logs, and exact-document validation

Publication

ICER
2024

Evaluating Exploratory Reading Groups for Supporting Undergraduate Research Pipelines in Computing

ICER · Computing education research · Published June 12, 2024

↗

04 / Get in touch

Have a hard problem
worth solving?

ashchembu@gmail.com ↗

Fremont, California · Bay Area

GitHubLinkedInRésumé

© Ashwin Chembu