Sukhbir Dhillon

About Sukhbir Dhillon

SummaryI am an artificial intelligence enthusiast who believes that it will continue to transform and improve our lives in the near future. Interested in the vast domain of machine learning and in my early career I specialized in computer vision and robotic perception using deep learning and probabilistic techniques. I have solid applied ML research and agile software development experience, having worked at unicorn startups both in the US and India, spanning high-performance C++ perception systems, edge deployment, and large-scale ML pipelines.Recently my work has expanded into ranking/recommendations and multimodal LLM/VLM systems, with a focus on evaluation, experimentation, and productionization at scale. The emphasis is on building end-to-end improvement loops - defining measurable quality signals, performing systematic error analysis, and using fine-tuning and distillation to translate those signals into deployable models. I enjoy the applied researcher sweet spot: rigorous research paired with production impact. My renewed focus is on developing and deploying multimodal AI applications on mobile, web and cloud platforms - while keeping a strong grounding in measurement, reliability, and production constraints. Knowledge Domains: • Ranking / Recommendations / Search: Learning to Rank (LTR), A/B testing, reranking, personalization, semantic/vector/knowledge-graph retrieval • Computer Vision: VQA, OCR, object detection (2D/3D), keypoint estimation, panoptic segmentation, multi-object tracking, SLAM/SfM/VIO • Generative AI: VAE/GAN/diffusion, txt2img/img2img/txt23d, multimodal embeddings, LLM/VLM, SFT/PEFT, RLHF/DPO/GRPO, knowledge distillation • Machine Learning: supervised/unsupervised, reinforcement learning, ablations + error analysis • Agentic AI: tool use, agentic workflows, structured outputs + validators, agentic memory, guardrails • Software / Systems: OOD, architecture, microservices, streaming/batch pipelines, reliability Tech Stack: • Languages: Python, Scala, Java, C++ • Modeling / Research: Jupyter/Colab, NumPy, Pandas, SciPy, scikit-learn, XGBoost, PyTorch/Lightning, Tensorflow/Keras, Deepspeed, Hugging Face, OpenCV, PCL • LLM / AI: vLLM, LangChain, LangGraph, Qdrant, Ollama • Data / Backend: PySpark, HDFS, Kafka, PostgreSQL, MongoDB, Redis, FastAPI, gRPC, REST, graphQL • MLOps / Infra: MLflow, Airflow, Docker, Kubernetes, Jenkins, Git/GitHub Actions • Deployment / Optimization: ONNX, TFLite, TensorRT, CMake, OpenMP/TBB • Cloud: AWS (S3, EC2, Lambda, API Gateway, ECS, SageMaker)

Experience

1. Applied Researcher

   eBay

   Aug 2025 — Present · Bengaluru, IN

   Working as an applied researcher on the recommendations team at eBay in GenAI based use cases for the e-commerce platform

Education

• Panjab University

  Master of Business Administration - MBA, Finance and Marketing

  2014 — 2015

• Panjab University

  Bachelor’s Degree, Mechanical Engineering

  2010 — 2014

• University of Florida

  Master’s Degree, Mechanical Engineering (Robotics)

Skills

• Statistical Modeling
• C++
• C
• Programming
• Systemverilog
• Tensorflow
• Optimal Estimation
• Embedded Systems
• Microsoft Office
• Industrial Robots
• Machine Learning
• Opencv
• Deep Learning
• Algorithms
• Linux
• Modelsim
• Html
• Matplotlib
• Xilinx Ise
• Corel Draw
• Python
• Mechatronics
• Autocad
• Scikit-Learn
• Rtos
• Embedded C
• Keras
• Matlab
• Agile Application Development
• Visual Studio
• Control Theory
• Ipopt
• Marketing Research
• Numpy
• Ansys
• Data Analysis
• Vhdl
• Simulations
• Autonomous Vehicles
• Project Management