About O-Health
O-Health is a health-technology company dedicated to democratizing and bridging the healthcare access gap across India and emerging markets. We are building intelligent digital health and AI systems that enable patients, doctors, and healthcare workers to communicate naturally through voice-based, multilingual interactions. Our technology supports timely and accurate healthcare, even in underserved regions.
Website: o-health.in
About the Role
We are looking for a Machine Learning Engineer – Edge, Vision, ASR & NLP who can own the end-to-end pipeline from raw camera/audio data to optimized models running on Android smartphones and Raspberry Pi.
You will work on:
- Vision models that use cameras for screening and health workflows
- Multilingual
ASR
(with Indic languages) to understand real patient speech NLP
models to extract medical data- Deploying all of this
on-device and at the edge
for real production application.
This is a highly hands-on role where one engineer stitches together ML research, edge deployment, and real-world healthcare impact.
What You Will Do
Vision & Camera ML
Build and refine computer vision models using live camera feeds (e.g., document/OCR, basic visual checks, camera feed cues).
- Design pipelines for capturing, preprocessing, and streaming camera data from Android tablets/phones and Raspberry Pi.
2.ASR for Indian Languages
Fine-tune and adapt ASR models (Whisper / wav2vec2 / NeMo etc.) for Hindi and other Indic languages
, handling noise, accents, and code-mixed speech.- Build streaming / chunked ASR pipelines that work reliably in real-world noisy clinical environments.
3.NLP for Medical Data extraction
Develop NLP models to extract medical data.
- Design pipelines for the applications.
4.Edge ML on Android & Raspberry Pi
Convert and optimize models (vision + ASR + NLP) to run on-device
using TFLite / ONNX Runtime- Work closely with Android and backend teams to integrate models into apps and microservices.
- Profile latency, memory, and power usage; apply quantization, pruning, distillation and other optimizations to make models production-ready on edge devices.
5.Data, Evaluation & Iteration
Help design and maintain datasets: multilingual audio, video, and text collected from deployments.
- Define and track metrics: WER/CER for ASR, precision/recall/F1 for NLP, FPS/latency for edge models.
- Iterate quickly based on real user feedback from PHCs, doctors, and field teams.
6.Collaboration & Ownership
Collaborate with product, design, ops, and medical advisors to translate field problems into ML tasks.
- Own features end-to-end: from idea → PoC → edge deployment → monitoring → improvement.
What You Will Need
1.Core ML & Coding Skills: Strong proficiency in Python
2.Solid hands-on experience in
- Computer Vision (OpenCV, CNNs, detection/segmentation/OCR etc.)
- ASR / Speech models (Whisper, wav2vec2, NeMo, ESPnet, etc.)
- NLP / Transformers
3.Edge ML & Systems Experience
Experience deploying ML models on Android
(TFLite / ML Kit / ONNX Runtime Mobile / PyTorch Mobile / ExecuTorch).- Experience running models on
Raspberry Pi or similar SBCs
with cameras/mics/sensors. - Understanding of edge constraints: CPU-only or small GPU, limited RAM, power limits, offline/spotty network.
- Familiarity with techniques like
quantization, model size reduction, lower input resolution, mixed precision
, etc.
4.Tooling & Infra
Comfortable working in Linux
environments, SSH, logs, basic networking.- Experience with
Git
, basic CI/CD and preferably Docker
or similar for packaging services. - Ability to expose models as
APIs
(FastAPI/Flask) and debug end-to-end flows. Language & Domain Comfort
Practical familiarity with at least one Indic language
(e.g., Hindi, Gujarati, Telugu, etc.) and willingness to work with noisy, code-mixed speech and regional accents.- Genuine interest in
healthcare / social impact
and sensitivity while handling patient data.
5.Mindset & Soft Skills
Ownership mentality: you enjoy taking things from idea → implementation → deployment → iteration.
- Comfortable with ambiguity, changing priorities, and cross-functional collaboration in a startup.
- Clear communication with both technical and non-technical stakeholders (doctors, field teams, partners).
real, deployable ML systems
Attributes We Value:
- Problem-solving mindset and ownership.
- Strong communication skills.
- Ability to work independently on-site.
- Willingness to learn and evolve quickly in a real-world healthcare environment, and work the extra mile.
Contact & Company Details
Company:
Address:
Email:
Website:
