Currently
Building local-first tools and production-grade agents.
The thread running through the work: make systems easier to reason about after they leave the prototype stage.
Selected work
Recent shipped surfaces
Local-first agent memory Tacit Captures coding sessions across Claude Code, Codex, and Gemini, then makes intent, decisions, blockers, and project history searchable locally. Session memory, local-first search, developer workflow capture. 
Career evidence system InChronicle Connects to daily work tools and turns overlooked contributions into structured career evidence for reviews, resumes, and job searches. Work capture, impact narrative, AI-assisted professional memory. 
Agentic AI security Intentyx Shipped agent observability, integrations, UI system work, and digital-twin UX for mapping real-world production systems into agentic workflows. MCP-style agent/tool integrations, security context, observability, digital-twin UX. Profile note
Context
I build distributed systems — the kind where a bad design decision at 3am shows up as a page six months later. I’ve set technical direction for systems processing 10B+ events/day, re-platformed batch pipelines into streaming architectures, and built a research product from zero to 150 paying users.
I write design docs before I write code. I debug by reading data, not guessing. I’d rather delete a system than add a flag to it. And I believe most production incidents are the quiet revenge of a shortcut someone took six months ago.
What I Do
I like two kinds of work: taking an idea from whiteboard to first production SLO, and taking a system that barely works and making it 10x faster or cheaper. I don’t build for CVs — I build systems that earn their keep, and I kill the ones that don’t.
My depth is in backend engineering (Go, Java, Scala, Python), polyglot datastores (Postgres, ClickHouse, Elasticsearch, DynamoDB, Redis), and data pipelines (Kafka, Spark, Airflow, Iceberg). I know enough cloud and SRE (AWS, GCP, SLOs, error budgets, blameless postmortems) to navigate — not to pretend it’s my craft.
Lately I’m building with AI and agentic systems — RAG pipelines, multi-agent orchestration, function calling, and the unglamorous work of making agents survive contact with real users: bounded tool calls, cost guardrails, and recovery paths for when the model is confidently wrong.
How I Work With AI
The engineering process has shifted in the last year, and I’ve built enough depth to make sure my work still clears the bar — maybe a notch higher than before. I run my own maker-checker-arbiter workflows across every step of the craft: high-level design, implementation, testing, documentation. No blind trust in the model.
My rule with AI: maximum ROI per hundred lines of code. No slop.
Other Engineering Projects
Multi-Broker Investment Platform
Unified portfolio tracking across Zerodha and Dhan. Built for personal use to consolidate holdings and P&L across Indian brokers. Currently inactive due to recent regulatory requirements around static IP fixation for broker API access.
Recent Work
Software Architect
Lana · OutOfTheBlue · Intentyx · Jul 2025 – Dec 2025
Architected a medallion data platform for Lana.Eco (Singapore) processing 25 years of public markets data across 5 countries — 1B+ data points, sub-second query latency on S3/Athena.
Delivered a production BigQuery-to-dashboard platform for OutOfTheBlue.ai in 6 weeks with a reusable framework adopted across multiple customers.
Shipped agent observability and integrations for Intentyx, plus the UI design system and digital-twin UX for modeling real-world production systems with agentic workflows.
Built 2 working prototypes for Product User Interface. 1, 2
Drove research on AI Agents and Observability. 1, 2, 3
Co-founder, Tech & Product
8Sapience · Apr 2024 – Jun 2025
Built an alternate research platform for equity analysts and investors with 150+ paying users. Shipped web app, search-grounded RAG, and Chrome extension. Built 70+ data pipelines processing 2K+ news articles daily from 50+ business sources with sentiment analysis and NER.
Principal Technical Architect
Splore · May 2023 – Apr 2024
Set technical direction for multi-tenant, terabyte-scale search platform across gaming and private equity sectors. Drove adoption of schema registry and semantic search across 3 product teams.
Technical Director, Engineering
Druva · Jan 2022 – May 2023
Led 12-engineer team on org-wide data lakehouse implementation. Replaced full-table batch reloads with incremental processing and resumable compute — eliminating the largest cost driver in DB-to-BI pipeline. Redirected team away from a Snowflake migration that would have cost 4 months of rework.
Earlier Highlights
- Finaxar — Re-architected card transaction engine (Scala, Akka, ClickHouse), doubling throughput with event-sourced processing and multi-node failover. Grew 6 engineers, 2 became technical leads.
- Symantec — Set technical direction for real-time event ingestion handling 10B+ events/day across 600+ nodes globally. Maintained 99.9% SLO.
- Adsparx (Acquired by Discovery Inc) — Scaled ad platform to millions of daily video impressions. Enhanced open-source open-m3u8 library for dynamic ad insertion, adopted by industry.
Education
B.E. Electronics from V.J.T.I. Mumbai (1999–2003)