Looking for a job

Amit User

@amituser13

Message

I build and validate robust computer vision, transformer, and machine learning systems.

India

Message

What I'm looking for

I'm looking to build and validate robust ML, LLM, and GenAI systems, with opportunities to work on model risk, explainability, fairness, monitoring, and responsible AI deployment.

At LAT Aerospace, I built a ViT/DETR object detector from scratch in PyTorch, reaching 0.85 mAP, and optimized production inference with TensorRT, FP16, and CUDA Graphs for a 2–3x speedup.

I design validation protocols around held-out data, precision-recall, robustness under distribution shift, failure-mode analysis, and documented model limitations. I also integrated ByteTrack and benchmarked a YOLO + DeepSORT baseline for multi-object tracking.

Previously, I developed real-time sensor-fusion and trajectory-prediction algorithms at Sharang Shakti and built a 3D Gaussian Splatting reconstruction pipeline at Samsung Research that achieved approximately 28 dB PSNR. I’m especially interested in responsible AI, model risk management, explainability, fairness, and robust LLM and GenAI deployment.

Experience

Work history, roles, and key accomplishments

Current

Machine Learning Engineer

Current

LAT Aerospace

Feb 2026 - Present (7 months)

Developed a Vision Transformer object detector from scratch in PyTorch, achieving 0.85 mAP. Designed evaluation protocols, optimized for production with TensorRT, and integrated ByteTrack for multi-object tracking.

PyTorch Vision Transformer Facebook DETR TensorRT FP16 CUDA Graphs ByteTrack YOLO Deep SORT Maps Precision Recall Explainability Model Monitoring

Software Engineer

Sharang Shakti Pvt. Ltd.

May 2025 - Feb 2026 (9 months)

Built real-time decision-making algorithms with sub-100 ms latency, integrating sensor fusion and trajectory prediction. Implemented data pipelines for high-rate sensor data and collaborated cross-functionally under Agile.

Sensor Fusion Trajectory Prediction Data Pipelines Agile Code Review Technical Documentation

Data Scientist

Samsung Research

May 2024 - Jul 2024 (2 months)

Implemented and trained a 3D scene reconstruction system using 3D Gaussian Splatting, achieving ~28 dB PSNR. Conducted ablation studies and assessed model behavior across downstream domains.

Gaussian Splats SFM PSNR Ablation Studies Hyperparameter Tuning Repeatability

Education

Degrees, certifications, and relevant coursework

Indian Institute of Technology Delhi

B.Tech, Mathematics and Computing

Graduated with a B.Tech in Mathematics and Computing in 2025. Coursework included Probability & Statistics, Linear Algebra, Machine Learning, Optimization, Algorithms, and Stochastic Processes.