|
I am a software engineer and AI/ML researcher working at the intersection of robotics, technology, and community. I hold an M.S. in Computer Software Engineering from Grand Canyon University and a B.S. in Computer Science from University of Silicon Valley. I currently work as an AI Robot Operator at Physical Intelligence in San Francisco, generating training data for general-purpose robotics foundation models. I am also the Founder and Executive Director of House of Seva, a California nonprofit serving the Punjabi and South Asian community of Solano County. I have experience as a Software Developer at Kaiser Permanente Regional Laboratory and as an AI Response Evaluator at Handshake AI. My research interests span AI/ML, healthcare informatics, and embodied intelligence. I am seeking full-time Software Engineering and AI/ML positions in 2026. Email / Resume / ResearchGate / LinkedIn / GitHub |
|
|
|
|
Jaiveer Bassi Edge AI, Model Optimization, Deployment Presented a comprehensive pipeline for efficiently packaging and deploying AI models for edge inference using quantization, pruning, lightweight containerization, and specialized runtime frameworks. Achieved up to 4× model size reduction and 50–70% faster inference on edge devices with minimal accuracy loss, validated on MobileNet and other models deployed on embedded GPUs and CPUs. DOI: 10.13140/RG.2.2.23010.59844 |
|
Jaiveer Bassi Deep Learning, Medical Imaging Built a reproducible ResNet-18-based classification pipeline for brain MRI scans using transfer learning and PyTorch. Classified brain tumors into four categories (glioma, meningioma, pituitary, healthy) from 7,000+ MRI images. DOI: 10.13140/RG.2.2.21638.28484 |
|
Jaiveer Bassi Transformers, LLMs, Code Generation Researched balancing speed, accuracy, and efficiency for locally deployed LLM code assistants on consumer hardware. DOI: 10.13140/RG.2.2.30622.86081 |
|
Jaiveer Bassi Adversarial AI, CNNs Analyzed cross-architecture adversarial attack transfer using FGSM, PGD, and CW on ResNet, VGG, and MobileNet (CIFAR-10). DOI: 10.13140/RG.2.2.16410.76489 |
|
|
|
CAN Bus Diagnostics Rig | C++, Python, SocketCAN, Embedded Systems Built a 3-node CAN bus on custom hardware and induced real bus-off, arbitration, and connector faults, diagnosing each via SocketCAN error-state analysis with a reproducible fault catalog. [Read on Medium] / [GitHub] |
|
GELLO | C++, Python, Arduino, Servo Control Rebuilt the core mechanism behind GELLO-style robot teleoperation for under $30, substituting potentiometer-based leader encoding for the servo-as-encoder approach used in production rigs, preserving direct joint-space control with no inverse kinematics. [Read on Medium] / [GitHub] / [GitHub] / [GitHub] / [GitHub] / [GitHub] / [GitHub] |
|
|
|
Python, RTDE, Docker, URSim Built a control and fault-injection harness against UR5e's production RTDE interface using UR's official simulator, reproducing a real segmentation fault from a production traceback and characterizing which connection failures are catchable in application code versus requiring external process supervision. Logged synchronized joint telemetry at fixed sample rate with automated gap detection to recover diagnostic evidence from failures that crash the process before any exception is raised. [GitHub] |
|
Python, Imitation Learning, Closed-Loop Policy Evaluation Built the codebase for published research quantifying how demonstration failure modes degrade closed-loop policy performance, directly informing how bad demonstrations propagate through imitation learning pipelines. [GitHub] |
|
Python, Labelbox, COCO JSON, Data Ops Data-ops pipeline for building a robot-arm grasp annotation dataset on Labelbox's free tier, exported to COCO JSON. Runs with no hardware. [GitHub] |
|
Python, Streamlit, ML, Anomaly Detection Built a synthetic robot fleet log analyzer featuring fault classification, recurrence detection, anomaly detection, and rule-based root cause analysis with a real-time Streamlit fleet health dashboard. Designed an end-to-end ML pipeline for processing structured log data across distributed robot systems, enabling automated triage and reporting without manual intervention. [GitHub] |
|
PyTorch, Deep Learning, Computer Vision, Matplotlib Implemented ResNet-18 from scratch and compared against a plain CNN to demonstrate why residual connections matter. Achieved 10% accuracy improvement through skip connections that solve the vanishing gradient problem. Includes comprehensive visualization suite with Grad-CAM attention maps, gradient flow analysis, and interactive Streamlit dashboard. [GitHub] |
|
Python, Docker, MLOps A Python library and CLI to package ML models into Docker images for fast inference deployment. Supports multiple model formats and optimized containerization. [GitHub] |
|
Python, Networking, VXLAN, OSPF Network simulation tool for VXLAN overlay networks with OSPF routing protocol implementation. Simulates spine-leaf fabric topology with VTEPs and VNIs. [GitHub] |
|
Python, NumPy, Matplotlib An interactive Python tool that trains a multi-layer perceptron on a synthetic shapes dataset and maps its inner workings over time. Visualizes neuron activations, weight magnitudes, training loss, accuracy, and forward passes through animated network diagrams using Matplotlib. [GitHub] |
|
PyTorch, Python, OpenCV, CUDA Designed a GUI tool to visualize adversarial attacks (FGSM, PGD, C&W, DeepFool) with real-time image comparisons, perturbation maps, and 3D surface plots. [GitHub] |
|
C++, PyTorch, Python, PyGame Built a self-play chess engine with a C++ backend and PyTorch network, featuring bitboard move generation and real-time GUI visualization. Implements minimax with alpha–beta pruning and neural network evaluation for position assessment. [GitHub] |
|
Streamlit, BioBERT, PubMed API, UMLS, Python Built a biomedical research platform with PubMed integration for querying abstracts, summarizing with BioBERT, and visualizing author–citation networks and keyword clusters. MedStract.net |
|
OpenAI, React, Node.js, Supabase, Tailwind, PDFKit AI-powered resume analyzer with GPT-based feedback, ATS scoring, skills matching, Supabase backend, Tailwind UI, and PDF export. CareerTuner.org |
|
Next.js 14, TypeScript, WebAssembly, Vercel, REST APIs Built an order management web app with Next.js 14 (App Router, TypeScript) featuring real-time STL file analysis via a custom WebAssembly/JavaScript parser to calculate volume, dimensions, and print cost client-side. Integrates Etsy REST API, EasyPost for shipping, and Resend for automated emails. Powers a $2K+/month e-commerce business with a fully automated order pipeline. castline.studio |
|
Swift, SwiftUI, AVAudioEngine, Swift Charts iOS tinnitus therapy app built from scratch with real-time audio synthesis using custom biquad IIR filters, pink/brown noise DSP, and a personalized notch filter tuned to the user's matched tinnitus pitch, with zero third-party dependencies. Includes guided habituation exercises and a progressive silence-exposure ladder grounded in Tinnitus Retraining Therapy principles, with Swift Charts tracking distress and awareness over time. [GitHub] |
|
|
|
Languages: Python, TypeScript, JavaScript, SQL, C++, Java, HTML/CSS Robotics & Systems: Linux (systemd, dmesg, journalctl), USB/serial debugging, CAN bus fundamentals, C, x86-64 Assembly, Bash scripting, distributed system diagnostics Frontend: React, Next.js, Tailwind CSS Backend: Node.js, Express, Flask, RESTful APIs, Docker, Git, CI/CD, Serverless AI/ML: OpenAI API, LangChain, PyTorch, HuggingFace, RAG Pipelines, Prompt Engineering Data & Cloud: PostgreSQL, MongoDB, Supabase, Vercel, AWS, Google Cloud Other: WebAssembly, OAuth 2.0, Pandas, NumPy, OpenCV |
|
|
|
HackerRank Software Engineer Certificate — Mar 2026 Cisco Network Support and Security — Sep 2025 IBM Artificial Intelligence — Aug 2025 Stanford AI/ML in Healthcare — Jun 2025 Google Project Management — Sep 2024 |