Pranoy Radhakrishnan | Machine Learning Engineer

Experience

Avataar AI March 2026 – Present

AI Engineer — Bangalore

Aqxle AI April 2025 – Feb 2026

AI Research Engineer — Bangalore

Developed custom agents for automated survey data processing with natural language interfaces.
Implemented automated EDA pipelines for faster, consistent analytics.

Accessia Apr 2024 – March 2025

Computer Vision Engineer — Bangalore

Optimized ML pipelines in C/C++ and CUDA.
Implemented structured pruning, post-quantization, and QAT.

First Sense Nov 2021 – Feb 2024

Computer Vision Engineer — Bangalore

Led CV development for animatable 3D Morphable Models (3DMMs).
Built vision enhancement solutions for virtual classrooms.
Implemented modified DETR for human pose estimation.

Neuroplex Aug 2018 – Jan 2021

Co-Founder & CV Engineer — Trivandrum

Built AI video analytics platform deployed across Indian Oil petrol pumps.
Deployed CV models on NVIDIA Jetson devices.
Worked on tracking, LPR, re-ID, segmentation, and action recognition.

Open Source & Personal Projects

Open Source Contributions

Transfusion-PyTorch — Implemented unconditional KV caching, matching null-token sequence length to prompts for correct CFG alignment, and adding classifier-free guidance.
Rectified Flow-PyTorch — Implemented the Unconstrained Alignment (UA) Flow implementation
Mimic-Video — Improved noise scheduling, VAE input normalization, and action decoding with proprioceptive masking.
MetaController — Fixed Switching Unit logic by incorporating the previous subgoal (z_t-1) for correct hierarchical control.
Pi-Zero-PyTorch — Fixed CFG RoPE alignment, temporal buffering, device compatibility, and unconditional pass robustness.

Personal Projects

Technical Skills

Languages: Python, C/C++, CUDA
Frameworks: PyTorch, TensorFlow, Keras, LangChain
ML Systems: Diffusion, Flow Matching, Transformers, Agentic Architectures
Deployment: ONNX, TensorRT
Databases: Neo4j, MongoDB, MySQL
Math: Linear Algebra, Multivariable Calculus, Probability & Statistics