Software Engineer · Video · ML · Systems

Jayant Singh Bhati

Building end-to-end video platforms, real-time inference pipelines, and backend microservices where frames meet silicon.

Get in touch GitHub LinkedIn

Experience

04 entries
Jul 2023 — Present Arcturus Business Solutions

Video Management System

Go · Python · C++ · PyTorch · CUDA · GStreamer · FFmpeg

  • Contributed to an end-to-end Video Management System (VMS), focusing on backend microservices and video analysis.
  • Developed microservices in Go and Python for video capture, storage, and retrieval from cameras, integrating FFmpeg for efficient processing.
  • Engineered MongoDB and Redis infrastructure for streamlined video data management and inter-microservice communication.
  • Enhanced video analysis pipelines using Python and C++, integrating ML models for advanced video processing.
  • Collaborated within a multidisciplinary team to achieve project milestones and communicate technical insights.
Mar 2023 — May 2023 Arcturus Business Solutions

Speed Detection & Tracking Pipeline

C++ · OpenCV · GStreamer · TensorFlow · PyTorch · DeepSort

  • Developed an end-to-end pipeline for speed detection at a client site.
  • Utilized C++ and OpenCV with GStreamer for efficient feed capturing from cameras.
  • Implemented YOLOv8 for accurate vehicle detection in captured frames.
  • Employed DeepSort for vehicle tracking and speed prediction.
  • Used Kalman filter for precise speed estimation and Cosine Metric learning for vehicle reidentification.
Dec 2022 — Jan 2023 Arcturus Business Solutions

Real-time Camera Feed Capture Pipeline

C++ · OpenCV · GStreamer · MinGW · CMake · Ctypes · Python

  • Developed a Camera Feed Capture pipeline using C++, OpenCV, and GStreamer to process feeds from multiple cameras.
  • Implemented dynamic camera initialization from JSON using OpenCV with GStreamer and Boost Library.
  • Utilized multi-threading and dynamic batching for efficient processing of large camera arrays.
  • Integrated the C++ pipeline with Python using Ctypes for real-time frame sharing.
  • Embedded the project with existing systems, ensuring smooth camera feed capture and inference.
Aug 2021 — Mar 2022 IISER Bhopal · Dr. Nirmal Ganguli

MS Thesis — Fermion Sign Problem via ML

Python · JAX · Flax · Jit

  • Explored Quantum Monte Carlo methods for stochastic simulations.
  • Developed ML algorithms for simulating probability distributions using JAX and Flax.
  • Simulated material quantum states with Markov chains, incorporating unsupervised learning and physics principles.

Education

Aug 2017 — Jun 2022 IISER Bhopal

BS-MS, Physics

Indian Institute of Science Education & Research, Bhopal

Computer Science

Data Structures & Algorithms, Data Science & ML, Quantum Computer Science, Discrete Mathematics, Operating Systems, Computer Organization, Cryptography

Mathematics

Linear Algebra, Multivariable Calculus, Probability & Statistics, Group Theory

Workshops

Machine Learning at NIT-H by Coding Blocks, Internet of Things at IIT-BHU by Innovians Tech

Skills

Languages

GoPythonC++

ML & Video

PyTorchTensorFlowOpenCVGStreamer FFmpegDeepStreamTritonCUDA

Tools & Frameworks

JAX / FlaxJitMongoDBRedis GitCMake

Achievements

GATE Physics 2731 AIR · 2022
JEE Advanced Cleared · 2017