Skip to content
View abubakarsiddik31's full-sized avatar
🥲
TODO: sleep
🥲
TODO: sleep

Block or report abubakarsiddik31

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. 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
abubakarsiddik31/README.md

Abu Bakar Siddik

Lead AI Engineer — agent orchestration, real-world tool integration, low-level LLM control.

Website · Writing · LinkedIn · X


I build agent systems that hold up in production: multi-step orchestration with explicit state, tool integrations that survive real users, and evaluation harnesses that measure behavior before it ships.

Previously I led the Core AI team at AskTuring.ai, where a document platform I architected scaled from 100 to 10,000+ concurrent users at ChatGPT-level latency, with a 98% reduction in ungrounded answers. I won the Google Nano Banana Hackathon 2025.

Now

  • Co-founder & Lead AI Engineer at an early-stage legal AI startup — agentic software for lawyers
  • MSc in Computer Science (ongoing, RUET) — research in LLM alignment

Selected work

  • Golem — agent framework: orchestration primitives, tool use, and recovery patterns.
  • Agent Bench — evaluation suite for measuring agent behavior across real tasks.
  • CareerKor — live AI career product: tailored documents, application tracking, and an interview coach.
  • Axiom Wiki — developer knowledge base with MCP (Model Context Protocol) support.
  • Skill Manager — developer tool for managing reusable AI skill definitions.
  • MagicSpin 360° — single image to interactive 360° rotation. Winner, Google Nano Banana Hackathon 2025.

Full portfolio: abubakarsiddik.site/projects

Writing

Production engineering essays from building LLM systems under real traffic.

All posts: abubakarsiddik.site/blog

Background

  • Co-founder & Lead AI Engineer — early-stage legal AI (2026–present)
  • Core RAG & AI Team Lead — AskTuring.ai (2025–2026)
  • Machine Learning Engineer — Sazim Tech: private, air-gapped enterprise AI (2023–2025)
  • ML Researcher & Engineer — Intelsense AI: Bengali ASR and conversational AI (2022–2023)

BSc in Mechatronics Engineering, RUET.

For AI agents

This profile is machine-readable: abubakarsiddik.site/llms.txt and a public MCP server at abubakarsiddik.site/api/mcp.


I care about correctness, reliability, and shipping AI that actually works.

Pinned Loading

  1. claude-skills-collection claude-skills-collection Public

    A curated collection of official and community-built Claude Skills – extend Anthropic's Claude with powerful, modular capabilities for productivity, creativity, coding, and more.

    1.1k 218

  2. llm-from-scratch llm-from-scratch Public

    Learn LLMs by building them from scratch in PyTorch: GPT, BPE tokenizer, BERT embeddings, SimCSE, then fine-tuning, quantization and deployment. Free book: https://abubakarsiddik31.github.io/llm-fr…

    Python

  3. golem golem Public

    Go-first framework for dependable AI agents: typed dependencies and outputs, explicit tools, composable models, observable runs

    Go 9

  4. axiom-wiki axiom-wiki Public

    Inspired by Andrej Karpathy's llm-wiki — the idea that LLMs should maintain a persistent, compounding wiki rather than re-derive answers from raw sources on every query. https://gist.github.com/kar…

    TypeScript 111 7

  5. zdr zdr Public

    30 1

  6. agent-bench agent-bench Public

    Python