RESUME

Abdullah Al Mamun, PhD

Sr. Machine Learning Engineer at Atlassian · Ex-Meta · 8+ Years

Permanent Resident (no visa sponsorship required) · San Francisco Bay Area, CA

  Download PDF   LinkedIn   Google Scholar   Contact

Professional Experience

Sr. Machine Learning Engineer

Atlassian · Mountain View, USA

Sep 2025 – Present

  • Replaced GPT-4o-mini with fine-tuned Llama-3.2 1B for Jira/Confluence search intent detection — $1M CAPEX savings/year.
  • Built RL-based fine-tuning pipeline for SMART Answer generation — 20% coverage improvement.
  • Engineered ML-based Jira intent classifier (replacing rule-based heuristics) with multilingual support for global Jira users — 17% increment in query success.

Member of Technical Staff

Aisera · Santa Clara, USA

Jun 2024 – Sep 2025

  • Built end-to-end scalable multi-agentic architecture for enterprise AI — driving ~$3.4M estimated ARR.
  • Fine-tuned LLaMA 3 and deployed on AWS — reducing $4M in CAPEX vs. GPT-4.
  • Integrated RAG into conversational AI, enhancing real-time semantic search by ~93%.

Machine Learning Engineer

Meta · Menlo Park, USA

May 2022 – Jun 2024

  • LLM in Ads Creation: Fine-tuned LLaMA 3 (SFT, RLHF, KV-Cache, 4D parallelism, quantization, distillation) — 13% CTR ↑, 6% CVR ↑, ~$497M iRev.
  • Personalized Ads Recommendations: AutoCA clusters for ad ranking via targeting relaxation — 0.1% iRev improvement, ~$109M.

Research Assistant (AI / ML)

ML & Data Analytics Lab (MLDAG), FIU · Miami, USA

Sep 2018 – Apr 2022

  • End-to-end development of interpretable deep learning models for early cancer detection and drug recommendation.

Research Assistant (ML)

Research & Innovation (RI), KFUPM · Saudi Arabia

Jan 2015 – Dec 2016

  • Implemented LSTM-based sentiment analysis tool for customer feedback — 98% accuracy.

Technical Skills

ML / AI
LLMs · Fine-tuning (LoRA, QLoRA, SFT) · RLHF, PPO, DPO · RAG · Agentic AI · Multi-Agent Systems · Transformers · Quantization · 4D Parallelism · KV-Cache · LangChain · LangGraph · Intent Classification · NL2JQL · Offline Eval Pipelines
Domain
Search Ranking (Jira / Confluence) · FPS Intent Detection · SAIN (Smart Answer Intelligence) · Ads Ranking · Recommendation Systems (RecSys) · Conversational AI · Knowledge Discovery
Engineering
Python · PyTorch · TEAMServe (ML model serving) · Statsig / Switcheroo (feature flags) · A/B Experimentation · Docker / Kubernetes · CI/CD · AWS · Databricks · Socrates (SQL)

Education

Certifications

Awards & Honors

For more on what I'm working on, see About or reach out.