Numbers
| 10x | trade allocation throughput at Goldman Sachs, on 80% less infrastructure |
| $10k+ | monthly losses prevented by a trade position calculator |
| 4 wk to 1 | expansion lead time at Swiggy, after moving capacity to slot level in 7 days |
| 5 | enterprise contracts closed by the PII engine at HeyMarvin |
| 10+ | enterprise customers unblocked by data archival |
AI tools I use: Claude Code, Superset.
Work
Zenarate Jul 2026 - now
Senior Software Engineer II
Career breakFeb - Jun 2026
HeyMarvin AI Nov 2024 - Jan 2026
Senior Software EngineerEquafin Corp
- Built a data archival system that unblocked sales to 10+ enterprise customers and kept 30+ organisations compliant on data sovereignty.
- Shipped a PII processing engine for surveys. It closed 5 enterprise contracts.
- Built the feature-flag and plan-control system behind tiered access and self-serve billing.
- Ran design reviews and mentored junior engineers. Onboarding time dropped by about 30%.
- Worked with Product to turn compliance requirements into a phased roadmap.
Swiggy May 2022 - Oct 2024
Software Engineer IIInstamart, Insanely Good, SuprDaily
The one I remember best: Insanely Good had a big marketing launch coming and needed new dark store slots within a week. Capacity was tracked per zone, which was too coarse to hand out individual delivery slots without overcommitting. I moved it to slot-level tracking in 7 days. Expansion lead time went from 4 weeks to 1, and the launch went out on schedule with no order failures.
- Delivered real-time demand shaping v2. Peak-hour order failures fell by about 1% absolute, worth roughly $1k+ GMV a day.
- Led backend for the Insanely Good pivot, including the high-availability presentation services it ran on.
- Integrated delivery tracking end to end while the product migrated into the main Swiggy app.
Goldman Sachs Jun 2018 - May 2022
Associate and AnalystIntern from Jun 2018 to Apr 2019
- Reworked trade allocation to handle 10x the throughput on 80% less infrastructure.
- Built a trade position open-close calculator that prevented $10,000+ in projected monthly losses.
- Set up Kong API Gateway for the Advisor Portal, putting auth, rate limiting and routing in one place.
- Automated regression testing for derivatives exchange connectivity. Release cycles went from 8 hours to 2.
- Tuned Kafka processors for real-time trade data over the FIX protocol.
Nirma University
B.Tech, Information Technology
Things I've made
Muffle - A browser extension that turns off the feed on X but keeps posting, replies and messages working. It also filters posts by keyword, shows who unfollowed you and makes bookmarks searchable.
- Chrome extension
- Bamboozled
- Joey's game show from Friends, playable in the browser. Spin the wheel, answer trivia about the show and race Chandler up the ladder. A new wheel every day, with a leaderboard.
- Browser game
Kharcha - A personal finance tracker, still in progress. Flutter app, FastAPI and Postgres backend, all self-hosted on EC2. The AI features run on my own Ollama model, so no spending data goes to a third party.
- Flutter, FastAPI, PostgreSQL, Ollama
- Tokamak trajectory planning
- University research on fusion reactors: an actor-critic RL agent that plans robotic arm paths inside a tokamak. Writeup, project page.
- Python, reinforcement learning
- Game-playing agent
- A deep Q-network that learned simple mobile games to roughly human level. I built it to see how reward shaping changes convergence when rewards are sparse.
- Python, DQN
- Price tracker
- Watches prices across 5+ shopping sites and sends an alert when one crosses a threshold.
- Node.js, MongoDB
Get in touch
Email is best: lovlinthakkar99@gmail.com. If you'd rather talk, book 30 minutes.
I'm most interested in backend-heavy work in fintech, infrastructure, or early-stage startups.