Aakash Mallik

Generalist Engineer

Lil bit about me

Currently at Wells Fargo building Debt Capital Market platforms, and FAFO-ing AI research stuff in my free time.

  • From research to real systems — I like understanding the fundamentals and building bottom-up.
  • Chasing the full AI stack end to end — not the MERN kind, but chip architecture and GPU programming through distributed training and efficient inference, edge devices included. A bit better every day.
  • Currently learning: Agentic AI systems,LLM posttraining and LLM serving infra
  • Previously: Second Brain pipelines at NeoSapien, classical ML + speech-to-text infra, and won SUI Overflow 2025.
  • Off-screen: Literature, Anime, Webtoons, and (irregularly) at the Gym.
  • Connect with me on LinkedIn or X or drop me an email at aakashmallik7777@gmail.com

Education

NIT Karnataka, Surathkal : B.Tech EEE, Minor in CSE · 8.07 CGPA · 2021-25

National Public School, Rajajinagar : CBSE Class XII · 97.6% · 2019-21

Presidency School, Nandini Layout : ICSE Class X · 98.88% · AIR 5 · 2019

Experience

Wells Fargo : Software Developer

Aug 2025 - Present

Building the New Issuance Trading platform with React & Redux, supporting Debt Capital Market transactions across production releases.

  • Building agentic workflows & MCPs for business-context-aware coding assistants.
  • Shipped an AI-assisted Github PR review system — cut review time by 30%.
  • Owning the migration from WAF & Selenium testing automation to Playwright.

NeoSapien : ML Developer Intern

Aug 2024 - Mar 2025

Implemented the Second Brain pipeline for the QA chatbot.

  • Used RAG, Self-Discover & Chain-of-Thought; Benchmarked, wrote Evals and optimized prompting, chunking, and embedding strategies.
  • Deployed retrieval/storage back-end with Firestore & Qdrant — latency down 46.5%.

Wells Fargo : SDE Intern

May 2024 - Jul 2024

Led a 3-member team in the Process Transformation group.

  • Delivered RPA solutions replacing manual workflows.
  • Built a skill-based task-matching platform for ~1200 employees — allotment time down ~98%.

Freelance & Contract

Setconnect - Kirloskar Ferrous Limited : ML Engineer

Data-science engagement to cut coke consumption by ~20 kg/T in the iron-making process.

  • Ran exploratory data analysis and predictive modelling on plant process and laboratory chemicaltest data.
  • Carried out an on-ground plant visit to map the end-to-end process.

Attenomics Lab : ML Engineer

Semantic scoring for high-volume scraped Tweets.

  • DistilBERT + triplet-loss pipeline — ~10% over baseline, ~$200/month cost cut.
  • Built a hybrid RAG system with multi-tiered retrieval for a brand marketing & insights chatbot.

OnTheRecord Technology : Founding Engineer

Founding engineer on BlueCat, a courtroom transcription desktop app.

  • Built speech-to-text model infrastructure.
  • Coded key frontend components.