Hello, I'm Prabhjit Singh Dhillon
I build intelligent systems and scalable applications using cutting-edge ML/AI and modern web technologies.
About Me
Cause I am awesome
Prabhjit Singh Dhillon
Machine Learning Engineer
My Journey
I'm a passionate Machine Learning Engineer and Full Stack Developer with a strong foundation in both theoretical and applied aspects of AI and software development. My journey in tech began during my undergraduate studies, where I excelled in dual majors of Computer Science and Applied Mathematics.
At Foundever, I built and deployed scalable ML models, optimized real-time detection pipelines, and designed end-to-end ML infrastructure. I'm particularly interested in computer vision, natural language processing, and the optimization of neural networks for real-world applications.
I'm also passionate about competitive programming and knowledge sharing, having qualified for ICPC regionals and helped teach data structures and algorithms to juniors as part of the Hoplite CS club.
hacker@dev:~
_
Education
Bachelor's in Computer Science and Applied Mathematics
San Jose State University, CGPA: 3.7/4.0
Work Experience
AI Engineer
FAM
October 2025 - Present
- Built a node-based generative design platform end to end (React, React Flow, Zustand) with 25+ composable nodes for interior and architectural visualization.
- Designed a Node.js/Express API unifying Gemini/Imagen, OpenAI, Claude, FLUX Pro, Stability AI and Meshy behind one interface, with request queueing, retry/backoff and SSRF protection.
- Shipped sketch-to-render, virtual staging, multi-style transfer, region-based editing, section/elevation-to-render and image-to-3D pipelines.
- Fine-tuned custom LoRA adapters on diffusion models using curated interior imagery, building the full dataset pipeline (curation, auto-captioning, trigger tagging) and tuning rank and learning rate against overfitting.
- Served trained LoRAs through a ComfyUI-based diffusion backend, letting designers stack brand-specific styles inside workflows without retraining the base model.
- Released the platform on web and iOS (Capacitor) and deployed it with PM2 on a VPS.
AI & Robotics Engineer
Whoopmasters India (Indian Army Defence Projects)
July 2026 - Present
- Built an AI aerial-surveillance system on DETR (Detection Transformer) for real-time object detection, deployed and flown on operational drones.
- Engineered a custom ground control station integrating ArduPilot over MAVLink, unifying mission planning, live telemetry, vehicle control and the AI detection feed in one operator interface.
- Developing ROS-based control software for an 18-servo hexapod robot, covering gait generation, IMU attitude sensing and failsafe disarm.
Machine Learning Engineer
Foundever
July 2023 - August 2025
- Built and deployed scalable ML models for recommendation and classification tasks, optimizing inference latency by 40%.
- Designed an end-to-end ML pipeline for large-scale data ingestion, preprocessing, and model deployment.
- Integrated DeepSORT tracking and YOLOv5 for multi-camera person tracking, optimizing real-time detection pipelines.
- Applied transfer learning and self-supervised learning techniques to improve feature extraction and model generalization.
- Supported Senior Staff: Prepared detailed project presentations and reports, aiding senior management in informed decision-making.
Achievements
ICPC Regional Qualifier
Click to learn more
ICPC Regional Qualifier
Secured college rank 2 and 297 in RL2 (2021). The International Collegiate Programming Contest is the premier global programming competition conducted by and for the world's universities.
Algorithms
Problem Solving
Competitive Coding
Hoplite CS Club
Click to learn more
Hoplite CS Club
Active member who taught Data Structures and Algorithms to juniors. Led weekly workshops and helped students improve their problem-solving skills and programming techniques.
Mentoring
Data Structures
Teaching
Hackathon - 3rd Prize
Click to learn more
Hackathon - 3rd Prize
Served as Tech Lead for a project to reduce food wastage. Developed an algorithm enabling local restaurants to donate food to homeless residents within a 5-mile radius.
Leadership
Hackathon
Social Impact
Skills & Technologies
Programming Languages
Deep Learning & Research
Generative AI
LLMs & Multimodal AI
Computer Vision
Training & Optimization
MLOps & Deployment
Full-Stack & Data
Featured Projects
Quantitative Stock Analysis Platform
March 2025 - PresentDeveloped a Flask API to deliver ML-driven stock predictions and financial insights. Built LightGBM models using RSI, MACD, and Bollinger Bands from Alpaca data. Optimized inference with Joblib for real-time performance.
View project
Multi-Camera Real-Time Tracking System
Dec 2024 - Feb 2025Implemented YOLOv5 and DeepSORT to detect and track subjects across multiple camera feeds simultaneously. Devised a mechanism to automatically close inactive feeds, optimizing resources. Reduced processing time by 30%.
View projectFull-Stack E-Commerce Platform
Aug - Dec 2024A comprehensive e-commerce solution with robust RESTful APIs, efficient MongoDB schemas, JWT authentication, and load balancing techniques for handling multiple simultaneous requests.
View project
Deeplense_gravitational-Images
Jan 2025 - March 2025The primary goal is to demonstrate whether incorporating physics-inspired architectural choices into the CNN design can lead to improved classification performance compared to a standard deep learning model (ResNet) and an ensemble approach for this specific astrophysics task. The comprehensive evaluation framework allows for a quantitative comparison of the different models.
View projectMini coinbase
April 2025 - presentDeveloped a full-stack cryptocurrency trading dashboard, simulating core Coinbase features like real-time price updates, portfolio balance display, and order placement. Utilized Go (Golang) for the backend API and WebSocket services, coupled with a React/TypeScript frontend. Implemented real-time data synchronization via WebSockets, feeding live market data into dynamic price charts built with Recharts. Constructed a responsive, component-based user interface using Material UI, managing application state effectively across components with Zustand.
View projectOpen Source Contributions
Many more contributions in different repositories, check my GitHub
DeepMind Torax(Google)
Contributed to DeePMind Torax model optimization library, built using JAX so simulations run quickly on GPUs/TPUs and can be easily plugged into machine learning models to help scientists figure out how to control the plasma better.
#884(Merged)
Enhanced ion heat source computation by integrating an absorption_fraction parameter and external power injection tracking, standardizing Q fusion calculations across modules.
#869(Merged)
Enhanced source model validation by blocking concurrent bremsstrahlung and Mavrin activation, eliminating double-counting of losses.
Neuroinformatics-unit/movement
Movement aims to provide a consistent modular interface to analyse pose tracks, allowing steps such as data cleaning, visualisation and motion quantification. We aim to support a range of pose estimation packages, along with 2D or 3D tracking of single or multiple animals.
#514
Implemented compute_straightness_index to measure trajectory complexity based on path straightness. Added unit tests and an example to validate and demonstrate its functionality. Updated package initialization to seamlessly expose the new function. .
#531
Added bidirectional option to filter_by_displacement, enabling displacement checks in both directions. Edge cases with oscillating points.
Gemini generative Ai
Contributed to Gemini(Generative ai Python And Genrative AI Js) support and documentation, added comprehensive documentation, and implemented pruning algorithms..
#409
Support Application Default Credentials
#401
Enhance generateContentStream with streamCallbacks support. Fixes #322.
#390
Add custom fetch and httpAgent support
#615
Add min_p #608 .
#614
[Ferature Request] Support Google ModelGarden Models #567.
#613
Add ToolConfig as Part of Live API Setup #558 Open.
Contact Me
Get In Touch
I'm always open to discussing new projects, creative ideas, or opportunities to be part of your visions. Feel free to reach out!
Follow Me & Stay Updated
Connect with me on social media for the latest updates, tutorials, and cool tech insights.