Artificial Intelligence Engineer

Location: Pune / Remote-Hybrid

Company: Sciverse Solutions Pvt. Ltd.

Experience Level: 45 years

Employment Type: Full-time

Department: AI & Advanced Technology

About Sciverse

Sciverse is a deep-tech MedTech company developing AI-powered diagnostic platforms and smart healthcare solutions. Our product suite includes point-of-care analyzers, AI-first microscopy systems, and intelligent software for image, text, and sensor data analysis. At the frontier of diagnostics and digital health, we engineer meaningful impact at scale.

Role Overview

We are seeking a highly motivated and skilled Deep Learning and Computer Vision Engineer to join our core AI team. You will design, train, and deploy models for microscopy image analysis, diagnostic automation, embedded vision systems, and contribute toward exploring vision-language models (VLMs) for structured report generation.

This is a hands-on engineering role where ownership, research depth, and deployment focus go hand-in-hand.

Key Responsibilities

1. Model Development & Training

Design and implement deep learning models for object detection, segmentation, classification, and metric learning.

Work with real-world biomedical and diagnostic datasets including microscopy, photometry, and point-of-care device outputs.

Apply image preprocessing, denoising, and augmentation pipelines to improve model robustness.

2. Model Optimization & Deployment

Optimize models for embedded and edge inference (using TensorRT, ONNX, OpenVINO, etc.).

Quantize and prune models for integration into diagnostic hardware platforms.

Collaborate with firmware and software teams for seamless deployment on custom devices.

3. Computer Vision Pipeline Engineering

Build scalable and reusable vision pipelines for tasks like cell counting, boundary detection, morphology estimation, and more.

Integrate multi-class detection and segmentation modules in no-code platforms like SciVision.

4. Vision-Language Exploration

Experiment with transformer-based models and VLMs (e.g., BLIP, Flamingo, LLaVA, SAM + LLM) for microscopy captioning, report generation, or structured interpretations.

Contribute to prompt design, fine-tuning, or hybrid modeling efforts.

5. Collaboration & Innovation

Work with cross-functional teams including biology, embedded systems, and UI/UX to convert scientific problems into engineering models.

Stay up to date with state-of-the-art methods in vision, multimodal AI, and biomedical imaging.

Must-Have Skills

Strong foundation in deep learning (CNNs, Transformers, Vision Backbones, etc.)

Experience with image classification, instance/semantic segmentation, object detection

Proficient in PyTorch and optionally TensorFlow

Experience with training on real-world noisy datasets

Understanding of model deployment, optimization (ONNX, TensorRT, TFLite), and performance profiling

Hands-on with OpenCV, NumPy, Pandas, and Scikit-learn

Strong debugging and visualization skills (e.g., TensorBoard, WandB)

Bonus / Nice-to-Have Skills

Familiarity with medical imaging (DICOM, TIFF, proprietary formats)

Exposure to multi-modal or vision-language models (e.g., CLIP, SAM, BLIP, LLaVA)

Knowledge of embedded vision (Jetson, Coral, RK3399) deployment

Publications or contributions to open-source in vision or biomedical AI

Familiarity with MLOps tools like MLflow, Docker, FastAPI

Who You Are

Self-starter with a strong sense of ownership

Passionate about solving real-world healthcare challenges with AI

Enjoys rapid prototyping, experimentation, and learning

Comfortable working in an agile, research-to-product environment

Why Join Sciverse

Work on mission-driven healthcare technology with tangible global impact

Solve unique AI and vision challenges at the intersection of biology, diagnostics, and embedded systems

Be part of a fast-growing, founder-led, engineering-driven team

Opportunity to take ideas from R&D to production in record time

To Apply

Email your CV, portfolio (if any), and a short note on why you're excited about this role to: [Confidential Information]