Role:
Senior Deep Learning Engineer (Core Model Training)
Location:
Bangalore (Hybrid)
Experience:
10 – 15 Years (Total)
The Opportunity
We are a Y Combinator-backed AI platform with over $40M in funding, revolutionising Intelligent Document Processing (IDP) . Our open-source models have nearly 1 million downloads, and our enterprise systems power 10,000+ global customers .
This is a rare, highly unique opportunity for a true builder. We are looking for someone to get their hands dirty building and training state-of-the-art foundation models from scratch. If you are currently solving high-scale ML challenges at global tech giants or deep-tech startups, this role offers the compute, autonomy, and mandate to push the boundaries of multi-modal AI.
Requirements
Who You Are & What You'll Do
-
True Hands-on Training: You have 5+ years of verifiable experience in pre-training, continued pre-training, SFT, PEFT/LoRA, or RLHF/RLAIF.
-
Modern Architectures: You work directly with Foundation Models like LLaMA, Qwen, Vision-Language Models (Qwen-VL, LLaVA), Vision Transformers (ViT), or LayoutLM.
-
Production-Scale ML: You optimize and deploy models for high-throughput inference using vLLM, TensorRT, Triton, or SGLang.
-
Hard Engineering Challenges: You will deploy large-scale multi-modal architectures, solve extreme table extraction edge cases (warped images, nested cells), and extract hierarchical information from complex document structures.
-
Measurable Depth: You can clearly articulate your GPU setups, distributed training strategies, dataset sizes, and concrete evaluation metrics (accuracy/latency/cost) on specific models.
Who You Are Not
-
You are not an API wrapper: Your core work is not limited to calling OpenAI, Claude, or Gemini APIs.
-
You are not a prompt engineer: You do not rely on prompt engineering as your primary ML problem-solving tool.
-
You are not just an orchestrator: You are not building your career solely on LangChain, agent orchestration, or basic RAG pipelines. You build the engines, not just the wrappers.
Core Requirements
-
10 to 15 years of total engineering experience, with 5-8 years specifically dedicated to core Deep Learning .
-
Undergraduate degree from an IIT or BITS Pilani is highly preferred .
-
Direct domain relevance in OCR, Document AI, multimodal learning, or Information Retrieval .
