Vowels of the People Association (VOPA) is a Pune‑based social‑tech organization that sits at the crossroads of artificial intelligence and education. Their flagship platform, NIPUN Maharashtra, powers foundational literacy and numeracy assessments for over 70,000 schools, touching millions of young learners across the state. VOPA’s mission is to democratise quality education by leveraging cutting‑edge AI solutions that work in low‑resource, multilingual environments, especially for Marathi and other Indic languages. The team is small but highly motivated, comprising researchers, engineers, and educators who collaborate closely to turn research ideas into production‑grade products that directly impact classroom outcomes. The AI/ML Engineer role is designed for fresh talent and early‑career professionals who can take ownership of end‑to‑end machine‑learning pipelines. You will be responsible for fine‑tuning open‑source large language models (LLMs) and vision models using LoRA, QLoRA, and PEFT, building robust computer‑vision pipelines for Marathi handwriting recognition, and creating speech‑recognition systems for low‑resource languages. The position demands a blend of research curiosity and production discipline – you will prototype, evaluate, and ship models on AWS, optimise inference latency with quantisation, ONNX and TensorRT, and integrate the solutions into FastAPI services that power VOPA’s educational products. Key Responsibilities: 1. Fine‑tune open‑source LLMs and vision models using LoRA, QLoRA, and PEFT frameworks. 2. Develop Marathi handwriting recognition and automated answer‑evaluation pipelines. 3. Build and optimise automatic speech‑recognition (ASR) systems for Marathi and other Indic languages. 4. Design end‑to‑end ML pipelines covering data ingestion, augmentation, training, validation and deployment. 5. Deploy models on AWS using Docker, ECS/EKS, and GPU inference endpoints; monitor cost and performance. 6. Apply model optimisation techniques such as quantisation, ONNX conversion, and TensorRT for low‑latency inference. 7. Conduct rigorous evaluation using metrics like Word Error Rate (WER) and Character Error Rate (CER); perform error analysis and iterative improvement. 8. Stay abreast of the latest AI research, implement promising ideas, and contribute to internal best‑practice libraries. 9. Collaborate with backend engineers to integrate AI services into FastAPI‑based micro‑services. 10. Document code, experiments and deployment procedures for knowledge sharing. Tech Stack: Python, PyTorch, Hugging Face Transformers, Unsloth, Docker, AWS (ECS, S3, Lambda, CloudFront), FastAPI, GPU clusters, ONNX, TensorRT, LoRA/QLoRA, PEFT, Git. Growth Path: Starting as an AI/ML Engineer, you can progress to Senior Engineer, Lead AI Engineer, or AI Product Manager, with opportunities to lead research projects, mentor junior talent, and influence product strategy. Why Join VOPA? You will work on AI solutions that directly improve education for millions, gain hands‑on experience with state‑of‑the‑art LLM fine‑tuning and low‑resource speech/vision AI, and be part of a purpose‑driven team that values learning, autonomy and social impact.