Portfolio

Rashaad N Mohammed

DATA & AI ANALYST · AUTONOMOUS AGENTS · BUILDER

Available for work · Bangalore, IN

I analyze complex data and build production-grade AI systems specializing in LLMs, multi-agent pipelines, and edge inference.

Currently Exploring Robotics & trying to give it tiny brain (LLM)

Download Resume Read the blog →
About
Who I am

I'm an AI & ML enthusiast, building things that actually work in production, fine-tuned LLMs, RAG pipelines, and distributed multi-agent systems designed for reliability at scale. My research has been published at IEEE CONECCT and ISCCSC.

I'm particularly drawn to the hard problems: AI evaluation, safety, and the gap between a model that performs well in a notebook and one that holds up in the real world.

📍 Bangalore, India 🎓 A.I.E.T · AI & ML · 2026 ✉️ rashaadnmohammed@gmail.com
Tech Stack

Languages

Python C Shell SQL

ML & AI

PyTorch TensorFlow Scikit-learn OpenCV HF Transformers AI Evaluation Quantization & Pruning

LLMs & Agentic Systems

Fine-tuning RAG Prompt Engineering Multi-Agent Architectures LangChain Ollama

Data & Databases

MongoDB Pinecone Pandas ETL Pipelines EDA Tableau Power BI

Infrastructure & MLOps

Linux Docker AWS (EC2 · S3 · Lambda) Cloudflare Tunnels GitHub / GitLab CI REST API Design Real-time Pipelines
Experience
Dec 2025 – Mar 2026 Freelance

Machine Learning Engineer

Startup · Bangalore

  • Trained ML models to classify web vulnerabilities, improving triage efficiency by 30% in automated pentesting workflows.
  • Engineered microservice architecture for real-time data pipelines in a distributed multi-agent system, reducing analysis time by 65%.
  • Fine-tuned LLMs on security datasets, achieving an 18% improvement in vulnerability classification and reducing false positives by 30%.
  • Built end-to-end ML pipeline from data collection to inference and reporting, containerized with Docker.

Publications

  • IEEE CONECCT · Jul 2025

    Fine-Tuning Techniques for Large Language Models: A Comprehensive Survey

    A systematic review of parameter-efficient fine-tuning methods LoRA, QLoRA, adapters, prefix tuning with comparative analysis of compute, memory, and downstream performance trade-offs.

  • ISCCSC · Dec 2025

    Enhanced Comparative Study of Summarization Methods for Legal Assistants

    Benchmarks extractive and abstractive summarization across legal corpora, identifying where domain-specific fine-tuning materially improves faithfulness over general-purpose LLMs.

Recent Writing
READ THE BLOG →
// TIMELINE

How I got here

  1. 2021

    A guy wanted business, so he sold me an Arduino

    That was the turning point for a bio student to fall in love with computers.

  2. 2022

    Built my first Bluetooth car controlled by phone

    With a 100-year-old Toshiba laptop, stealing the neighbour's WiFi and mind you, there was no ChatGPT.

  3. 2023

    Got my first laptop and chaos began

    Having no idea what to do, dummy me learned 3D Max, Revit, everything except programming.

  4. 2024

    Who in the hell introduced me to AI

    Built my first project, a recommendation system. Trust me, I spent more time debugging it. End result: it went on to do lexical search.

  5. 2025

    Not the CV (Computer Vision)

    My first hackathon realised not everything runs in production. Can't forget the embarrassment in Mysuru.

  6. 2026

    Now

    Graduated, learned AI, kept up with the latest tech, taught AI to multiple students, fine-tuned countless models.

// NOW

What I'm into right now

Last updated May 2026

Building
  • Multi-agent pentesting pipeline
  • Fine-tuning LLMs for specific tasks
  • This portfolio (apparently)
Reading
  • Think and Grow Rich (Napoleon Hill)
  • Artificial Intelligence: A Modern Approach (Stuart Russell)
  • Anything related to politics
Listening to
  • Nope podcast (Kunal Kamra)
  • NetworkChuck (networking)
  • Self-help podcasts
// MANIFESTO

Things I believe

  • You are on the right path if the bug is irritating you.
  • RAG is mostly retrieval. The "G" is the easy part.
  • AI agrees with you. That's the product, not your validation.
  • I measure worth by output value.
  • Stop blaming hardware for latency in chunk retrieval.
  • Know your terminal better than your best friend.
// SCARS

Things that didn't work

The failures are where the lessons live.

  • Tried to fine-tune Llama on Google Colab free tier

    Lost all checkpoints. Learned the hard way about resource limits.

  • Built a RAG system for a simple CRM

    Answers were great, but latency killed it. Later realised not everything requires AI, some tasks can use regex and get the job done.

  • Entered a hackathon with a complete project, failed to run on their env

    Library compatibility across OS is a real pain.

// INFLUENCES

What shaped how I think

  • Prashant Kishor's lens on data Founder, IPAC

    Came across him a few months back. Opened my eyes to what data can actually do far beyond the dashboards.

  • Rahul Patil CTO, Anthropic

    Whenever scaling problems hit, his name comes up. Built a reputation around being the person who solves them.

  • Attention Is All You Need Vaswani et al.

    Eight pages that ate the world. Every time I read it, there's something new to learn.

  • A Small Number of Samples Can Poison LLMs Anthropic & Alan Turing Institute

    Made me realize how brittle LLM safety actually is, a few hundred poisoned samples is all it takes.

// GLIMPSES

Behind the scenes

Fine-tuning a model
Fine-tuning a model
Whiteboard sketch
Whiteboard · thinking out loud
Terminal nmap session
nmap, my comfort show
Arduino system architecture
Arduino · system architecture
Teaching AI
Teaching AI to students