Saksham Arora

Saksham Arora

Data scientist, quant researcher and AI engineer. Known online as nerfsaksham.

Open to internships and full-time roles

Download résumé

At a glance

Education
BS in Data Science at IIT Madras and B.Tech in Computer Science at GGSIPU, both since 2024, in parallel.
Experience
Founding data engineer at Cimplifie since October 2024, with internships in data science at Airtel Xtelify and AI engineering at Tieout.
Research
Credit risk modelling, extending a multi-credit Lévy model by Prof. Graeme Baker (Columbia University), since August 2026.
Competition
AlphaNova Global Quant Competition 5: rank 6 of 6,000+. IMC Prosperity 4: rank 154 of 18,800+ teams, top 0.8%. Amazon ML Challenge 2026: rank 902 of 33,000+ teams.
Hackathons
Won the Paytm GenAI Hackathon and the Elastic x Google Cloud Hacknight in Delhi. Finalist at NASA Space Apps.

Selected work

Margin

Sits beside the lecture you are watching on Udemy, Coursera, DeepLearning.AI or YouTube, keeps its slides, follows what is said, and writes the study notes into your Obsidian vault with the code run. Then the crux, flashcards, and answers from the lecture itself.

Python, FastAPI, Chrome Extension, Obsidian

Visit the siteSource code

Latch

Pins a YouTube lecture in full screen so Escape does nothing and leaving takes a deliberate hold. Ships with an enforcement layer that closes the usual escape hatches, new tabs, extension toggles, popup dismissal.

JavaScript, Chrome MV3, Fullscreen API

Visit the siteSource code
ECX: everything claude codex, the Claude Code and Codex setup

ECX

My whole Claude Code and Codex setup, open source. Both agents share one memory, the vault answers from my phone with the laptop shut, and a doctor checks every hook, index and sync. 17 skills, 21 hooks, tests in CI.

Python, Claude Code, Codex, MCP, Hooks, ChromaDB, LightRAG, CLIP

Source code
HFT matching engine benchmark output

HFT matching engine

High-performance limit order book engine achieving 2.7M ops/sec with p99 latency of 900ns, a 3.31x improvement over a mutex baseline. Implements price-time priority matching, lock-free SPSC ring buffer, and 21 correctness tests.

C++20, Lock-Free, SPSC Ring Buffer, Order Book, CMake

Source code

See all 38 projects

Timeline

  1. Credit Risk Research, Building on Prof. Graeme Baker's work, Columbia UniversityExtending Baker and Capponi's multi-credit model (elastically stopped Lévy processes, arXiv:2608.10321) to firm-specific loadings on its common jump factor, with Ishaan Harish and Akshat Bhaskar.
  2. Founding Data Engineer, Cimplifie Tech SolutionsBuilding ETL pipelines and automation tools that cut manual data scraping by 70%.
  3. B.Tech Student, Computer Science & Engineering, GGSIPUBachelor of Technology in Computer Science and Engineering at Akhilesh Das Gupta Institute of Professional Studies, alongside the IIT Madras BS.
  4. B.S. Student, Data Science, IIT MadrasEnrolled in the IIT Madras Online BS programme in Programming and Data Science.
  5. Rank 902 of 33,000+ teams, Amazon ML Challenge 2026Business entity resolution: matching noisy business records to reference entities across US and India data and an unseen France split.
  6. AI Engineer Intern, TieoutAI engineering internship at Tieout, an agentic-AI startup automating specialized financial audits (Employee Benefit Plan audits) for CPA firms.
  7. Selected Attendee, OpenAI DevDay 2026Invited to OpenAI's annual invitation-only developer conference, selected from a global applicant pool.
  8. Data Science Intern, Airtel XtelifyData science internship at one of India's largest telecom operators.
  9. Winner, Elastic x Google Cloud Hacknight, DelhiBuilt Vayu Watch, a real-time Delhi air quality and public health platform on live DPCC/CPCB ground station data, with Gemini-generated personalized advisories and an Elasticsearch-indexed live history.
  10. Winner, GenAI Hackathon (Paytm)Led the multi-agent FastAPI backend (6-agent orchestration, 28 endpoints) for CommuneOS, an AI-native community operating system, built in 48 hours.
  11. Quantitative Researcher, AlphaNova Global Quant Competition 5Finished rank 6 of 6,000+ registered data scientists.
  12. Quantitative Trader, IMC Prosperity 4Competed in IMC Prosperity 4 trading competition against 18,800+ teams globally.
  13. Finalist, NASA Space Apps ChallengeBuilt orbit-ops, a forecasting tool, as a finalist entry in NASA's global Space Apps Challenge.
  14. Systems Engineer, HFT Matching EngineBuilt a high-performance HFT matching engine in C++ achieving 2.7M operations per second with p99 latency of 900ns, a 3.31x improvement over a mutex baseline.

Hiring for data science, quant research or AI engineering? Email me.

Download résumé Open the full site
Quick answers

Saksham Arora is a data scientist, quant researcher and AI engineer from India. He goes by nerfsaksham on GitHub, X, Hugging Face, Kaggle, YouTube, dev.to, Instagram, npm and Topmate, and publishes at saksham.digital.

I am a data scientist and quant researcher from India, studying at IIT Madras (Online BS, Programming and Data Science). I build ML pipelines, quantitative trading models, and data-driven systems, and placed top ~0.8% globally at IMC Prosperity 4 out of 18,800+ teams. I also have a systems engineering background in low-latency C++ (lock-free data structures, HFT-style matching engines). I care about rigorous analysis, market microstructure, and tools that do one job extremely well.

Who is Saksham Arora?

Saksham Arora (nerfsaksham) is a data scientist, quant researcher and AI engineer from India. He studies at IIT Madras (BS in Data Science) and GGSIPU (B.Tech in Computer Science) in parallel, builds Margin, Latch and Claude Code mods, and writes at saksham.digital/blog.

Is nerfsaksham Saksham Arora?

Yes. nerfsaksham is Saksham Arora's handle on GitHub, X, Hugging Face, Kaggle, YouTube, dev.to and other sites. His own site is saksham.digital.

What has Saksham Arora built?

Margin, a Chrome extension that writes study notes into Obsidian while you watch a lecture; Latch, which pins a lecture or PDF in full screen so Escape does nothing; four Claude Code mods (teach-me, vhs, frugal and lofi) and the awesome-claude-mods list; LMEX, an hourly LLM latency benchmark; and a C++ limit order book matching engine benchmarked at 2.72M orders per second.

Where does Saksham Arora write?

On his blog Chimera at saksham.digital/blog, mostly about AI model launches read through the numbers and the tools he builds. Some posts are also on dev.to/nerfsaksham.

How can I contact Saksham Arora?

By email at saksham10arora@gmail.com, or on X as @nerfsaksham.

Every profile

npx saksham-arora opens an interactive card in a terminal, and curl saksham.digital returns the same thing as text.

Download résumé