ASAtul Sharma
Say hello ↗

A little more about me

An engineer.
A person, too.

I’m Atul. I build systems, make things useful, and like figuring out how they work.

At AlgoTest, I build systems for researching, testing, and running quantitative trading strategies. I like being involved from the first idea to something people can actually use. At The Streamliners, I worked on MVPs and multiple projects, trying different ideas to find what worked. I care about the whole product and what happens after it ships.

Numbers help me see whether the work is making a difference. Seeing that impact is a big part of what keeps me going.

I ask a lot of questions. If I don’t understand something, I want to break it down until I can explain it simply. Sometimes that means reading a paper. Sometimes it means building a small version myself and seeing where it breaks.

That curiosity takes me from trading systems and databases to ML, LLM agents, and tools I wish existed. My side projects are a place to follow those questions and try things out. I’m also interested in learning how language models work from the ground up.

I write about what I’m learning, when I have something to share.

Away from code, I like tennis, running, and reading. This is my little corner of the web, with room for life, observations, and whatever I’m curious about next.

Instagram ↗My writing ↗
Atul Sharma smiling in a plaid shirt
Beyond the keyboardTennis, running & reading.

The journey

Products, platforms, production.

Feb 2025 – PresentAlgoTestY CombinatorS22Founding EngineerQuantitative trading infrastructure and LLM agent workflows.See moreSee less

What I did

  • Lead engineering across signals, distributed backtesting, and trade execution, building systems that let traders research, test, and run custom strategies over years of derivatives data and billions of daily market events.
  • Build low-latency C++ market-data pipelines, real-time signal processing, and 24/7 execution systems, balancing latency, throughput, data correctness, and recoverability.
  • Own architecture, delivery, and production reliability across the Execution and Signals teams, improving fault isolation, recovery, debugging, and performance as the platform scales.
  • Worked on LLM agents that operate the website, with human review at multiple stages, an agent harness for workflow orchestration, and Model Context Protocol (MCP) tools to access platform data.

Tech stack

C++PythonOpenOnloadUDP multicastPolarsNumPyLokiLLM agentsAgent harnessMCP
Jan 2024 – Jan 2025HackerRankY CombinatorS11Software EngineerML systems and backend platforms for assessment integrity.See moreSee less

What I did

  • Worked on the Integrity Team, building ML-powered assessment systems spanning plagiarism detection for LLM-generated code, webcam proctoring, and liveness detection.
  • Architected a PII-compliant identity verification system and its backend services, designing privacy and security boundaries around candidate identity data.
  • Improved ML plagiarism pipelines and Elasticsearch data reliability to support production assessments at scale.
  • Spearheaded engineering efforts for bias audits on HackerRank’s automated employment decision tools (AEDT).

Tech stack

Machine learningOptunaElasticsearchPythonRuby on RailsAWS
Jun 2021 – Dec 2023The StreamlinersFounding EngineerLocatIN, Karobar Easy, and MVPs for early-stage startups.See moreSee less

What I did

  • Built and scaled LocatIN, an indoor maps and navigation product that helps people find their way through large infrastructure facilities.
  • Led development of Karobar Easy, a comprehensive business management solution for small and medium-sized enterprises.
  • Delivered MVP prototypes for early-stage startups, helping them validate ideas and accelerate growth.

Tech stack

Android SDKKotlinJavaNode.jsMongoDBFirebase