Senior Data Scientist ( Kernel Optimisation & Inference Engineer)
Eka CareWhitefield, BangalorePosted 2 months agoA free account shows the apply link for every role and matches you to the ones that fit your resume.
About the role
EkaCare is a connected healthcare platform with medical models used in production. You will optimize GPU training and inference, quantize smaller models, and enable deployment on clinic and mobile hardware.
What you'll do
- Profile training and inference workloads to identify utilization gaps
- Write and tune CUDA kernels when existing operations limit performance
- Optimize MoE grouped GEMMs, communication, and expert load balancing
- Build fast inference paths with batching, caching, and speculative decoding
- Own quantization for 2B and 4B model variants and verify evaluation parity
- Enable on-device inference on hardware used in Indian clinics
Skills
Must have
- CUDA
- GPU performance optimization
- Kernel optimization
- Model inference
- Quantization
- Profiling
- Memory hierarchy optimization
- Moe optimization
Nice to have
- Fp8
- Torch.compile
- On-device inference
- Mobile inference
Tools
- Vllm
- Tensorrt-llm
- Sglang
- Flashattention
- Awq
- Gptq
- Inductor
Who it's for
- Experience
- 2 to 5 years
- Domain
- healthtech
About Eka Care
Eka Care has 19 open roles in Bangalore, at Eka Care, Windmills of your mind (EPIP). See every Eka Care role.
More at Eka Care
Mid level · On-site · 4+ years
Mid level · On-site · 3 to 5 years
Internship · On-site · 0 to 1 years · Fresher friendly
Mid level · On-site · 2 to 3 years
Entry level · 1 to 3 years · Fresher friendly
Mid level · On-site · 3 to 5 years
Mid level · 4 to 8 years
Entry level · 1 to 3 years · Fresher friendly
Similar roles
All ml and ai engineering jobsEli Lilly and Company · Manager · 10+ years
