united states / ai engineeropen-to-work

now-shipping

I'm an AI engineer with a decade+ building production systems in financial services — now designing GenAI and agentic systems end to end, from RAG and eval harnesses to the CI that ships them.

See what I've shipped Play commit-streak

move your cursor here ↘

newai systems2026

Agent Memory

Durable cross-repo memory for Claude Code, GitHub Copilot, and Codex. Each client engagement is a separate store rather than a filtered view, so one engagement's notes are unreachable from another even by exact id. Knowledge exports at global scope by default — what you learned travels between machines, the client's architecture does not. Zero runtime and zero dev dependencies, asserted in CI and published to npm with Sigstore provenance attestation.

Runtime deps

Tests

CI jobs

open to new opportunitiespythonfastapiragagentic aievalslanggraphlitellmqdrantneo4jopensearchawskuberneteskafkamcp

utkarsh.about()

A decade+ of backend. Now building agents.

I'm an AI engineer with a decade+ building production backend systems in financial services — trading, money movement, and securities management — including four years embedded at Fidelity Investments as an EY consultant, delivering event-driven microservices and high-traffic investment platforms.

Now I design and ship GenAI and agentic systems end to end: tool-calling agents, multi-step orchestration, and RAG with hybrid retrieval across dense, BM25, and graph signals. I build the evaluation harness that gates promotion — golden datasets, versioned baselines, regression gates — so every retrieval and model change is measured before it ships. I also lead AI enablement for 200+ engineers across the institutional vertical.

toolkit/

Languages: Python, TypeScript, Go, Java, SQL
GenAI & Agents: RAG, Hybrid retrieval, LangGraph, CrewAI, MCP, LiteLLM, OpenAI SDK
Evaluation: Golden datasets, Regression gates, Guardrails, PII-aware handling
Backend: FastAPI, asyncio, Kafka, SSE streaming, Microservices
Infra & Data: AWS, Kubernetes, Terraform, PostgreSQL, Qdrant, Neo4j, OpenSearch

utkarsh.work()

selected-work

AI Systems2026

npm Package

Agent Memory

Node.js · node:sqlite · Agent Skills

Runtime deps

Tests

CI jobs

Sigstore

Provenance

AI Systems2025

RAG Platform

Code Analyzer

A code-intelligence platform that ingests and queries large multi-repository codebases — 600+ repos at its first engagement — to power software-engineering audits. A tri-store backbone pairs Qdrant for dense semantic search, Neo4j for call graphs and symbol references, and OpenSearch for BM25 exact-match on the rare identifier tokens embeddings handle worst, fused with Reciprocal Rank Fusion and cross-encoder reranking.

Python · Qdrant · Neo4j · OpenSearch · LiteLLM

600+

Repositories

73%

Keyword recall

520ms

p95 latency

Retrieval stores

AI Systems2026

AI Agent

Costco Price Match

An AI agent that scans Costco receipts, cross-references each line item against active deals and temporary price drops, and emails a weekly price-adjustment report. Receipt parsing runs on Amazon Nova; the weekly run is an AgentCore Runtime triggered by EventBridge. Ships with a native SwiftUI companion app — CostScanner — on the App Store.

AWS Lambda · Amazon Nova · AgentCore · SwiftUI

Nova

Receipt parsing

Weekly

Run cadence

AgentCore

Runtime

App Store

Companion

Apps2026

iOS App

Yantra: Second Brain

A native iOS app for capturing voice memos, notes, journals, meeting recordings, photos, and saved URLs — then chatting with them. Transcription, embeddings, and summarization all run on-device via Apple Speech, NLEmbedding, and iOS 26 Foundation Models; only the retrieved snippets and the question leave, sent under the user's own Anthropic key through a stateless proxy that persists nothing. Captures English, Hindi, and Hinglish with translation.

SwiftUI · Apple Speech · Foundation Models · Anthropic

Languages

100%

On-device NLP

iOS 26

Foundation Models

Bytes retained

utkarsh.experience()

experience

Building the retrieval platform behind engineering audits, and the agents that automate enterprise onboarding.

Built Code-Analyzer, a code-intelligence platform that makes large commercial financial-services codebases auditable on a consulting timeline, treating retrieval over code as a problem distinct from retrieval over prose.
Built the evaluation harness that gates promotion — 30 hand-curated Q&A pairs across 15 indexed repositories, scored on keyword recall, file-pattern hit rate, cross-module recall, and citation rate against a versioned baseline. Hybrid retrieval holds 73% keyword recall at 520 ms, with 100% of answers citing sources.
Designed the TerraCore Onboarding Agent — an agentic system driving Terraform, Jenkins, and account-servicing APIs as LLM tools, cutting application onboarding from weeks to hours with resumable, auditable workflow state.
Shipped the agent as an async FastAPI service with SSE progress streaming, and made destructive infrastructure operations safe to automate through input validation, allow-listed actions, dry-run modes, and human-in-the-loop approval gates.
Leads AI enablement for 200+ engineers across the institutional vertical, covering agentic frameworks and org-scale GitHub Copilot adoption.

Led event-driven platforms and cloud-native modernization for Fidelity Institutional.

Led design and delivery of an event-driven microservices platform for Fidelity Institutional's Custom Separately Managed Accounts, launching with 200–500 high-value institutional accounts representing roughly $1B in assets under management.
Delivered a cloud-native Mainframe Analysis Tool that automatically maps dependencies across thousands of modules spanning COBOL, JCL, PL/I, CICS, IMS, and DLI — cutting retirement-impact analysis from weeks to minutes and replacing manual lookups for analysts.
Architected distributed workflows across AWS Lambda, EventBridge, and SQS for trade allocations, securities servicing, and cash movements; provisioned environments with Terraform and hardened data integrity with JSON schema validation and standardized git hooks.

Modernized Fidelity's high-traffic retail investing platform.

Modernized Fidelity's retail investing platform across the Periodic Investment Plan, baskets, and the 52-Week Challenge, letting consumers automatically invest from external bank accounts into mutual funds, ETFs, and equities on weekly, bi-weekly, and monthly schedules.
Built multi-step workflow orchestration on AWS Step Functions for periodic cash transfers, securities purchases, and fund allocations.
Automated infrastructure with Terraform, added CI/CD, and instrumented Datadog logging, tracing, metrics, and alerting for high-volume investment flows.

Built resilient banking and insurance systems for a regulated financial exchange.

Built and supported banking and insurance applications for a Texas Department of Insurance-regulated financial exchange, owning critical production incidents and 24x7 batch processing.
Automated incident handling and alerting with Python and Shell, feeding an Angular monitoring dashboard; built reconciliation jobs to maintain data accuracy across enterprise member-data utilities.
Led offshore teams in Manila and Chennai and achieved the lowest incident 'pain minutes' metric in the organization; delivered across COBOL, PL/I, JCL, Java, Python, and Datastage, including IMS to UDB migration pipelines.

/ Education

2019 — 2020M.S., Computer ScienceUniversity of Texas at San Antonio

2009 — 2013B.E., Computer Science and EngineeringVisvesvaraya Technological University

/ Certifications

Databricks Accredited Generative AI Fundamentals
Developing Serverless Solutions on AWS

utkarsh.play()

commit-streak

Squares light up and fade. Tap them before they go out to commit them to the graph. Faster taps land as darker greens. Miss three and the run is over.

contributions: 000
streak: 00
best: 000
lives: ■■■

git commit --often

Fill the graph.

lessmore

utkarsh.writing()

notes-from-building

Posts

Medium

Your Coding Agent Has a Memory Problem. Storing More Won’t Fix It.

Aug 8, 202610 min read

Four rules I built into a memory system for Copilot and Claude Code, after concluding that a confidently wrong note is worse than no note at all. Open a second window. Your agent knows…

aigithub-copilotsoftware-engineeringnodejs

Medium

LLM evals: the complete field guide

Jul 26, 202624 min read

Here’s a confession it took me years of writing software to even notice: I have almost never had to wonder whether my code’s output was correct. Something always told me. The compiler…

large-language-modelsartificial-intelligencemachine-learninggenerative-ai-tools

Medium

I built a second brain that runs on my iPhone — and the cloud does almost nothing

May 31, 20267 min read

I wanted a notes app I could talk to. Not a chatbot stapled to a text box. A real place where everything I capture during the day — a voice memo on a walk, a meeting, a journal entry, a…

generative-ai-toolsproductivitysecond-brainios

Medium

Nine Seconds to Zero

Apr 28, 20265 min read

A credential mismatch in a staging environment is small, ordinary friction. You notice it, you fix the config, you move on. On Friday afternoon, the AI coding agent running at PocketOS, an…

claude-opusdbdeleteai-agentincident

Medium

Most developers at work use GitHub Copilot.

Apr 6, 20265 min read

Most developers at work use GitHub Copilot. The best learning resources were built for something else. There’s a gap that nobody talks about, and I want to close it. At my company, we have…

co-pilotgithub-copilotaiclaude

Let's build

something good.

Have a project, a role, or an idea worth chasing? My inbox is open — I read everything and reply to most.

now-shipping

Agent Memory

A decade+ of backend. Now building agents.

selected-work

Agent Memory

Code Analyzer

Resume Screener

Costco Price Match

Yantra: Second Brain

Passport Tool iOS

Meeting AI

Pull Vids

Flaunt GitHub

Always Decimal

Everyday Developer Tools

Feast Factor

Copilot Anatomy

Copilot How To

experience

commit-streak

notes-from-building

Your Coding Agent Has a Memory Problem. Storing More Won’t Fix It.

LLM evals: the complete field guide

I built a second brain that runs on my iPhone — and the cloud does almost nothing

Nine Seconds to Zero

Most developers at work use GitHub Copilot.

something good.

now-shipping

Agent Memory

A decade+ of backend. Now building agents.

selected-workselected-work

Agent MemoryAgent MemoryAgent Memory

Code AnalyzerCode AnalyzerCode Analyzer

Resume ScreenerResume ScreenerResume Screener

Costco Price MatchCostco Price MatchCostco Price Match

Yantra: Second BrainYantra: Second BrainYantra: Second Brain

Passport Tool iOSPassport Tool iOSPassport Tool iOS

Meeting AIMeeting AIMeeting AI

Pull VidsPull VidsPull Vids

Flaunt GitHubFlaunt GitHubFlaunt GitHub

Always DecimalAlways DecimalAlways Decimal

Everyday Developer ToolsEveryday Developer ToolsEveryday Developer Tools

Feast FactorFeast FactorFeast Factor

Copilot AnatomyCopilot AnatomyCopilot Anatomy

Copilot How ToCopilot How ToCopilot How To

experienceexperience

Aug 2025 — PresentLead Engineer, AI & PlatformEY · Aug 2025 — Present

Jan 2023 — Aug 2025Lead EngineerEY · Fidelity Investments · Jan 2023 — Aug 2025

Oct 2021 — Jan 2023Senior Software EngineerEY · Fidelity Investments · Oct 2021 — Jan 2023

Mar 2021 — Sep 2021Full-Stack DeveloperEY · Capital Group · Mar 2021 — Sep 2021

May 2020 — Mar 2021Cloud Engineer (Intern)Healthcare Triangle Inc. · May 2020 — Mar 2021

Jul 2014 — Nov 2018Software EngineerHCL Technologies · Jul 2014 — Nov 2018

commit-streak

notes-from-buildingnotes-from-building

Your Coding Agent Has a Memory Problem. Storing More Won’t Fix It.Your Coding Agent Has a Memory Problem. Storing More Won’t Fix It.Your Coding Agent Has a Memory Problem. Storing More Won’t Fix It.

LLM evals: the complete field guideLLM evals: the complete field guideLLM evals: the complete field guide

I built a second brain that runs on my iPhone — and the cloud does almost nothingI built a second brain that runs on my iPhone — and the cloud does almost nothingI built a second brain that runs on my iPhone — and the cloud does almost nothing

Nine Seconds to ZeroNine Seconds to ZeroNine Seconds to Zero

Most developers at work use GitHub Copilot.Most developers at work use GitHub Copilot.Most developers at work use GitHub Copilot.

something good.

selected-work

Agent Memory

Code Analyzer

Resume Screener

Costco Price Match

Yantra: Second Brain

Passport Tool iOS

Meeting AI

Pull Vids

Flaunt GitHub

Always Decimal

Everyday Developer Tools

Feast Factor

Copilot Anatomy

Copilot How To

experience

notes-from-building

Your Coding Agent Has a Memory Problem. Storing More Won’t Fix It.

LLM evals: the complete field guide

I built a second brain that runs on my iPhone — and the cloud does almost nothing

Nine Seconds to Zero

Most developers at work use GitHub Copilot.