GenAI - Data Scientist

Overview About Ripik.AI

Ripik.ai is a fast-growing industrial AI SAAS start-up founded by IIT D/ BITS alumni and with extensive experience in McKinsey, IBM, Google and others. It is backed by marquee VC funds like Accel, Venture Highway and 25+ illustrious angels including 14 unicorn founders.

Ripik.ai builds patented full stack software for automation of decision making on the factory floor. Today, they are deployed at more than 15 of the largest and most prestigious enterprises in India including the market leaders in steel, aluminium, cement, pharma, paints, consumer goods and others.

It is one of India’s very few AI product start-ups to be a partner to GCP, Azure and AWS. We are also the AI partner of choice for CII, ICC and NASSCOM.

About The Role

We’re building an agentic AI platform that turns one line of text and a video feed into end-to-end, real-time computer-vision solutions—think semantic video search, object / action recognition, and task-oriented visual agents deployable with a single click. As a Gen AI ML Engineer, you’ll architect the core vision & multimodal-reasoning stack and pave the road from prototype to production.

Roles And Responsibilities

Semantic video search

Ship a pipeline that allows users to type “show every forklift near aisle 5 in the last 30 minutes” and get keyed-off clips in

Wire embeddings to a hybrid FAISS/HNSW index; surface results through a simple REST & React playground.

Create agentic pipelines

• Chain vision language models and zero/few-shot vision models with LLM planners (Gemini, GPT-4o, AutoGen, etc.) so a single prompt becomes a multi-step perception workflow.
• Profile and accelerate inference (TensorRT, ONNX, quantization, batching) to meet latency / throughput targets on GPU and CPU fleets.
  
  

Rapid prototyping loops

Run weekly “paper-to-prototype” spikes: reproduce a fresh arXiv idea, benchmark, and decide go/no-go in < 5 days.

Hand successful python scripts & checkpoints to MLOps for productionization—no plumbing marathons.

Data & Evaluation

• Spin up scalable pipelines for video ingestion, labeling (active learning, weak supervision), experiment tracking, and continuous evaluation.
  
  

Collaborate & Lead

• Partner with product and ML Ops engineers; set research direction, mentor future hires, and establish best practices.
  
  

Must-have Skill Set

• 1–3 years deep-learning research experience (internships & grad work count).
• Fluency in Python + PyTorch; comfortable hacking large vision/LLM repos.
• Proof you ship ideas—first-author paper, OSS repo, Kaggle medal, or faithful reproduction of a cutting-edge model.
• Hands-on with LLM prompting/fine-tuning and at least one agent framework.
• Able to turn fuzzy product asks into measurable experiments and explain results clearly.
  
  

Bonus Cred

• Large-scale video retrieval or temporal grounding experience.
• Prior work building agentic-AI pipelines that combine perception models with LLM reasoning.
• Open-source contributions to GenAI/vision libs (OpenCLIP, Vid2Seq, ViperGPT, etc.).
  
  

What can you expect?

• Ability to shape the future of manufacturing by leveraging best-in-class AI and software; we are a unique organization with niche skill set that you would also develop while working with us
• World class work culture, coaching and development
• Mentoring from highly experienced leadership from world class companies (refer to Ripik.AI website for details)
• International exposure
  
  

Work Location - NOIDA (Work from Office)