atharvajoshi01 (Atharva Joshi) · GitHub
Skip to content
View atharvajoshi01's full-sized avatar
🎯
Focusing
🎯
Focusing
17 followers · 8 following

Achievements

Achievement: Pull Shark

Achievements

Achievement: Pull Shark

Highlights

  • Pro

Block or report atharvajoshi01

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
atharvajoshi01/README.md
Atharva Joshi field notes header

I work on machine learning systems that need to hold up in the real world.

That usually means some mix of regulation, latency, and reliability.

Portfolio • LinkedIn • Email • New York

What I Work On

Area What pulls me in
AI governance explainability, audit trails, fairness checks, policy-aware tooling
Production RAG hybrid retrieval, ensemble generation, evaluation that maps to a real decision
Quant systems order books, pricing, execution research, market microstructure
Agent evaluation measuring accuracy, latency, cost, and safety before something ships

I am less interested in building one more polished demo and more interested in building systems that survive constraints.

Selected Work

Project What it is Live
finreg-ml Governance-oriented ML wrapper for explainability, fairness, drift, threshold tuning, and EU AI Act compliance reporting. On PyPI. demo
talent-rag Production-grade RAG: hybrid retrieval (semantic + metadata + keyword), cross-encoder reranking, ensemble generation with judge selection, async index-build pattern. repo
stocksense End-to-end demand forecasting with walk-forward CV and a reorder-quality decision metric instead of MAPE alone. dashboard
clarify LLM agent that turns a free-text BA brief into a validated artifact pack (requirements, RACI, traceability) without hallucinating. repo
crypto-stat-arb Cointegration stat-arb on real Kraken data with Kalman hedge ratios and explicit gross vs. net Sharpe analysis. repo
agenteval LLM-agent evaluation framework. Comparative against a deterministic baseline, so the deployment decision is one number. repo
Atlas C++20 limit order book engine. 62M ops/sec, 16ns insertion, NASDAQ ITCH 5.0 parsing at 258M msg/sec, lock-free ring buffers. repo

How I Like To Build

  • start from the failure modes, not the press release
  • keep the workflow clear enough that another engineer can audit it
  • prefer evidence over posturing
  • treat performance and correctness as product features

Open Source Work

These are actual public PRs. Merged contributions get verifiably attributed on the upstream commit log.

Merged

Open and under maintainer review

Right Now

  • making regulated ML workflows easier to explain and review
  • treating agent evaluation like engineering work instead of theater
  • moving closer to systems where implementation detail matters

Recent Activity

Recent commits

  • no recent personal-repo commits picked up

Recent open source PRs

  • #328 stefan-jansen/zipline-reloaded Fix DataPortal correctness bugs: tuple vs None and wrong frequency closed 2026-10-15
  • #512 ranaroussi/quantstats Add compounded flag to calmar() and rar() closed 2026-09-26
  • #103 joshyattridge/smart-money-concepts Fix look-ahead bias in swing_highs_lows open 2026-09-21
  • #1359 kernc/backtesting.py Fix read-only array error in FractionalBacktest indicator scaling merged 2026-07-20
  • #364 bukosabino/ta Add Rank and Percentile indicators open 2026-06-08

Recently published repos

  • eligible Patient-to-trial matching on harmonized clinical data: OMOP harmoniza… 2026-07-10
  • ab-test-decision-framework Streamlit web app that turns raw A/B test CSVs into a ship/no-ship de… 2026-06-27
  • nyc-tlc-congestion-pricing-causal-study Causal (difference-in-differences) study of NYC congestion pricing's… 2026-06-26
  • talent-rag Production-grade RAG system for talent intelligence — hybrid retrieva… 2026-06-05
  • stocksense End-to-end demand forecasting and inventory health platform for a dis… 2026-05-13

last updated 2026-10-21 UTC

Contact

Pinned Loading

  1. mms-fault-classification mms-fault-classification Public

    Vibration fault classification dashboard

    Python 1

  2. agenteval agenteval Public

    Lightweight evaluation framework for AI agents — measure accuracy, cost, latency, and safety across any agent architecture

    Python

  3. finreg-ml finreg-ml Public

    Regulation-aware ML pipeline for finance — explainability, fairness audits, and EU AI Act compliance built into model training

    Python

  4. Atlas Atlas Public

    Low-Latency Order Book Engine with Predictive Execution - C++ core with Python ML pipeline

    Python

  5. crypto-stat-arb crypto-stat-arb Public

    Statistical arbitrage engine for cryptocurrency markets — cointegration-based pairs trading with walk-forward backtesting

    Python

  6. talent-rag talent-rag Public

    Production-grade RAG system for talent intelligence — hybrid retrieval, cross-encoder reranking, ensemble generation, FAISS, FastAPI, Streamlit

    Python