Nikhil Mahajan

01.

About Me

I'm an enthusiastic problem solver who thrives on tackling hard, ambiguous challenges. Give me a blurry roadmap, and I’ll bring clarity through execution—figuring things out as I go. I’ve worked in fast-paced startup environments under tight deadlines, where adaptability, ownership, and speed were key.

Over the years, I’ve worn multiple hats—engineering, product, team management, and even customer interaction. I’ve led small teams, collaborated across diverse functions, and know how to stitch together the moving parts into a coherent solution. I’m passionate about building real, impactful products and love being involved end-to-end—from brainstorming ideas and giving product feedback to writing code and gathering insights directly from users.

For me, it’s not just about writing great code—it's about contributing to something meaningful and continuously pushing the product and team forward.

02.

Experience

Astrikos.AI

Software Engineer (Full Time)

July 2025 – Present | Bangalore

Engineered domain-specific conversational AI chatbots by fine-tuning open-source LLMs and developing backend infrastructures for diverse enterprise clients.
Deployed models across multi-GPU environments, implementing model partitioning techniques to overcome single-node memory constraints.

Astrikos.AI

Software Engineer (Intern)

April 2024 – June 2025 | Remote

Designed and developed a flow engine to enable dynamic execution of Python code through a node-based system.
Built a dynamic console for managing global datacenter operations, improving visibility and control.

Recepto.AI

Software Engineer (Intern)

Nov 2024 – March 2025 | Remote

Designed and implemented a workflow to regularly retrieve the latest and most relevant leads from the internet.
Developed custom web scraping tools to extract leads from multiple online sources...

Repello.AI

Software Engineer (Intern)

March 2024 – May 2024 | Remote

Designed and developed the user interface for an automated red teaming platform, based on a thorough understanding of product needs.
Built the platform with features including real-time log streaming...

Wikimedia Foundation

Google Summer of Code

May 2023 – August 2023 | Remote

Developed end-to-end (E2E) tests for various user journeys to improve application reliability.
Contributed to a large-scale codebase by fixing minor bugs in Wikipedia.

Self-Employed

Full Stack Developer (Freelance)

Nov 2022 – Dec 2022 | IIT Roorkee

Developed a web-based platform for simulating stock trading.
Integrated third-party APIs to fetch and display real-time stock market trends and data.

03.

Some Things I've Built

redis-server.c

Redis From Scratch

Built a lightweight, in-memory key-value database from scratch in C, replicating core Redis functionality and underlying data structures. Engineered a concurrent TCP server using standard C socket programming...

train_vae.py

Variational Autoencoder

Trained a custom Variational Autoencoder (VAE) from scratch on the Google QuickDraw dataset to autonomously generate and interpolate hand-drawn sketches. Designed an advanced encoder-decoder architecture...

gemma_finetune.ipynb

Bhagavad Gita AI

Curated a custom, domain-specific dataset on the Bhagavad Gita from scratch by scraping websites, running OCR on scanned books, and transcribing YouTube audio. Leveraged the Gemini API...

agent_graph.py

Shopmate

Engineered an AI-powered personal shopping assistant using a multi-step LangGraph agent to automate product discovery, autonomously comparing items and recommending the best options...

vote_secure.sh

Election Portal

Developed and maintained the official platform for conducting campus-wide elections at IIT Roorkee, enabling over 10K voters to elect Positions of Responsibility (PORs)...

cert_gen.py

IKS Console

Developed for the Indian Knowledge Systems (IKS) Division, Ministry of Education, Government of India. Built a dynamic certificate generation platform that allows users to upload templates...

Education

Indian Institute of Technology Roorkee

B.Tech in Chemical Engineering

Nov 2021 - April 2025

Other Noteworthy Activities

Project Manager - MDG Space

April 2022 – April 2025

Technical Club

Led multiple technical projects end-to-end from ideation to deployment.
Organized and conducted workshops, mentorship programs, and tech events attended by over 100 students. Delivered technical lectures on various tech stacks including frontend and backend development, and cloud. Mentored juniors for hackathons, internships, and technical problem-solving.
Fostered a collaborative and growth-oriented team culture.
Contributed to strategic discussions on integrating emerging technologies into club initiatives.

Participant - Google Gen AI Hackathon

Sep 2024 – Oct 2024

Developed a GenAI-powered application to support cancer diagnosis and treatment by analyzing MRI scans, identifying and segmenting tumors, estimating cancer probability, and generating detailed diagnostic reports with personalized treatment plans. Included a GenAI chatbot for real-time access to up-to-date research and treatment insights.
Collaborated in a 3-member team; selected among the top 5 teams nationally.
Invited to Google’s Gurgaon office; networked with VCs and participants, and received swag and recognition.

View Project Repo

04.

Skills & Stats

Languages

Python TypeScript C C++

Software Dev

React Next.js Django FastAPI PostgreSQL TimescaleDB MongoDB Redis Celery Selenium

GenAI

LLM Fine-tuning LangGraph Gemini APIs Video Generation VAEs Latent Diffusion Transformers

DevOps

AWS EC2 ECS ECR S3 Lambda SQS WAF Load Balancer Cloudwatch Terraform Docker Git Linux Nginx

05.

What I've Written

How does Node Js work behind the scene?

Node Js is not a programming language but it is a Javascript runtime environment. It is built on top of V8 Engine and LibUv. Node Js has the Event-Driven Architecture and the event loop is said to be the heart of this architecture.

Read on Medium →

Bits on the Move

Files are collections of binary digits. Whether it’s an image, audio, video, text, or any other data type, the interpretation of those binary values determines what the data represents. Binary data might seem a bit strange, but it’s the backbone of our digital world.

Read on Medium →

Guide to train on multiple GPUs using DDP

A simple, hands-on guide to understanding Data Parallelism. This post walks through a runnable script to show exactly how distributed training works under the hood—handling synchronization, seeding, and how GPUs actually communicate.

Read on Reddit →

Variational Autoencoder (VAE): How to train and inference?

Ever wondered how Variational Autoencoders work? This post breaks down the architecture, the math behind the training process, and how they generate new data. It also covers practical details I learned while building one myself.

Read on Reddit →

Overview of Wan 2.1: Text to Video Model

I spent some time deconstructing the inference pipeline of the Wan 2.1 model. Here is a step-by-step breakdown of how it takes a simple text prompt and transforms it into a full video using its T5 encoder, Latent DiT, and VAE decoder.

Read on Reddit →