Jishnu P
Deep Learning Scientist
Hire this person
Send a job offer directly to this candidate
About
Deep Learning Scientist with 4.5+ years driving production-grade Generative AI, multimodal AI, and end-to-end Computer Vision pipelines for video/image processing. Published researcher (PLOS One) with patents in AI-driven healthcare CV, specializing in GANs, diffusion models, Transformers, lip-sync systems, 3D U-Net segmentation. Proven expertise owning model training/inference orchestration, deployment, and bridging research-to-production for high-impact solutions.
Experience
Deep Learning Scientist – Video Dubbing(AI LipSync) & Localization - Cynapto Technologies - Mumbai, IN
(2024-04 - 2026-06)
- Built state-of-the-art lip-sync models for multilingual dubbing, aligning mouth movements with diverse audio tracks.
- Leveraged GAN and Diffusion-based synthesis with Transformer encoders to improve sync accuracy and realism by 25%.
- Optimized training/inference pipelines via PyTorch DDP orchestration on distributed HPC ,parallelizing tasks, managing dependencies, and slashing training time/inference costs by 30%.
AI Research Engineer – Life Science R&I Lab - Tata Consultancy Services (TCS) - Bengaluru, IN
(2022-06 - 2024-01)
- Developed "Automatic foot ulcer segmentation using conditional generative adversarial network (AFSegGAN): A wound management system", having direct impact on the chronic wound treatment which is $96.8B in United States.
- Created a conceptual model for 3D Heart segmentation using 3D-Unet, outperforming state-of-the-art research metrics.
- Developed GAN based model to segment the colon regions from the CT/MRI images (DICOM, NIfTI).
- Collaborated with clinical scientists, health systems, and academic universities to drive advanced R&D projects, ensuring validation for model safety and efficacy.
Machine Learning Engineer - Tata Consultancy Services (TCS) - Bengaluru, IN
(2021-06 - 2022-06)
- Developed and maintained LLM-ready/RAG-based document intelligence pipelines on Azure Cognitive Search, including data ingestion, indexing, and retrieval optimization.
- Processed and classified 1.5 million multi-format documents (PDF, XML, JSON) for business unit labeling.
Education
M.Tech - Artificial Intelligence & Data Science - Amrita Vishwa Vidyapeetham (2019-06 - 2021-06)
B.Tech - Computer Science & Engineering - Vidya Academy of Science & Technology (2013-06 - 2017-06)
