Lead ML/AI Engineer

Job Title: Lead, ML/ AI Engineer Location: Bangalore Reports to : AI Director About Digital Green Digital Green is a pioneer global not for profit organization, utilizing digital platforms and community-driven approaches to amplify the voices of smallholder farmers and improve their livelihoods. Our mission is to create a world where farmers use technology and data to build prosperous communities. By harnessing the power of technology, we facilitate knowledge sharing, capacity building, and market linkages, enabling farmers to adopt sustainable agricultural practices and increase their productivity and income. We are dedicated to transforming the lives of under-served smallholder farmers worldwide through innovative technology solutions. Backed by leading philanthropic organizations such as the Bill & Melinda Gates Foundation (BMGF), Walmart Foundation, USAID, and UK Foreign, Commonwealth & Development Office (UKFCDO), we are committed to leveraging data and technology to empower smallholder farmers and strengthen agricultural extension systems. Job Summary LLMs · Embeddings · Agentic Flows · Computer Vision · Audio Processing · Generative AI · Rapid Prototyping The Lead ML/AI Engineer will lead the development of high-reliability LLM and generative AI systems for Digital Green's farmer-focused platforms. We are seeking a Lead ML/AI Engineer with deep experience in LLMs, generative AI, and agentic architectures to design and deploy end-to-end production systems at scale. You will build optimized LLM inference pipelines, implement RAG and embedding-based workflows, and work with multimodal data (text, audio, image) to serve farmer-facing advisory tools. In this role, you will also define and run robust evaluation frameworks, ensure model safety and reliability, and operationalize production-grade ML pipelines with monitoring and continuous improvement. You will collaborate closely with product, engineering, and field teams to build AI systems that perform effectively in multilingual, low-resource, and real-world agricultural contexts. Key Responsibilities: LLM Development, Optimization & Evaluation - Fine-tune, adapt, and optimize large language models for domain-specific use cases (summarization, reasoning, conversational agents, generative tasks). - Design, test, and refine prompts using advanced techniques (CoT, in-context learning, few-shot prompting, agent orchestration). - Build internal prompt libraries, reusable templates, and evaluation datasets. - Develop robust evaluation frameworks for accuracy, coherence, hallucination rate, safety, bias, and reasoning ability - Automate evaluation pipelines to validate model updates and fine-tuned variants. Agentic Workflows & Agent Evaluation - Architect and implement agent-based systems with tool use, memory, planning, and multi-step task execution (e.g., LangChain, MCP-based systems). - Validate agent behavior through simulation of real-world workflows—handling tool calls, vector DB retrieval, error recovery, and escalation logic. - Stress-test agents in ambiguous or multi-turn scenarios for robustness, determinism, and reliability. - Build monitoring tools for agent behavior, performance drift, and anomalous actions. Automatic Speech Recognition (ASR) for Multilingual Agricultural Use Cases - Build and evaluate ASR systems optimized for agricultural domain content and rural speech patterns. - Work with targeted ASR models such as Whisper, Azure Cognitive Services, Speechmatics, or Wav2Vec-based systems. - Develop pipelines for multilingual transcription, diarization, noise handling, and accent robustness. - Validate transcription accuracy with field recordings, user-submitted audio, and noisy environments. Computer Vision Models for Agricultural Imagery - Develop and evaluate image models for crop detection, disease/pest classification, and agri-specific object identification. - Build image enhancement and preprocessing techniques for real-world farmer-submitted images (low light, blur, noise). - Work with CLIP, Vision Transformers, YOLO, Mask R-CNN, or DINO models for mobile-friendly and production-grade deployments. Embeddings, Retrieval & Knowledge Systems - Deep technical understanding of modern embedding models (e.g., Sentence Transformers, E5, GTE, Nomic, Cohere embedding families) and their behavior across languages, domains, and modalities. - Build and fine-tune agriculture-specific embedding models optimized for farmer speech, rural terminology, crops, pests, soils, weather, and region-specific agricultural practices. - Create multilingual embedding pipelines for Indian languages (e.g., Hindi, Telugu, Kannada, Bengali, Odia), handling code-mixing, transliteration, dialectal variations, and noisy field data. - Evaluate embedding quality using custom relevance metrics, clustering tests, semantic similarity benchmarks, and domain-specific retrieval tasks. - Design RAG workflows powered by fine-tuned embeddings, ensuring improved context recall, precision, and diversity. Security, Safety & Compliance - Test LLMs and agents against vulnerabilities including prompt injection, adversarial inputs, data leakage, and sandbox escape attempts. - Collaborate with security teams to ensure compliance with data privacy frameworks (GDPR, HIPAA, etc.). - Implement guardrails, safety evaluators, and content filtering systems. Cross-Functional Collaboration & Mentorship - Translate product requirements into model tasks, prompts, and evaluation strategies. - Mentor junior engineers in LLM evaluation, fine-tuning, embeddings, and agentic systems. - Partner with backend, DevOps, and product teams to deliver production-ready features and prototypes. Qualifications & Skills: - Education: Bachelor's, Master's, or Ph.D. degree in Computer Science, Engineering, or a related field with a focus on machine learning and NLP - Experience: Proven experience of 6+ years as a AI/ ML engineer, data scientist, or software engineer, with hands-on experience in deploying machine learning models - Deep expertise with LLMs, embeddings, RAG workflows, and agentic architectures. - Experience with ASR models (Whisper, Azure Speech), multilingual audio processing. - Hands-on experience with computer vision for classification, detection, and enhancement tasks. - Proficiency with Python, Transformers, LangChain, SentenceTransformers. - Strong understanding of guardrails, evaluation harnesses, and safety tooling. - Experience with RLHF, LoRA, quantization, distillation. - Prior experience building internal tools for evaluation, testing, or data pipelines. - Contributions to open-source LLM frameworks or multimodal systems. - Experience deploying models with Triton, TorchServe, TensorRT, or similar. - Team Player: Strong interpersonal and communication skills, with the ability to work effectively in a collaborative team environment and contribute to a culture of innovation and excellence What You'll Build: - Reliable, high-accuracy LLM systems powering real-world products. - Multi-modal systems involving Speech Recognition, Image Verification, Enhancement and Classification for real world applications - Autonomous agents capable of tool use, planning, and multi-step decision making. - Scalable RAG pipelines with multimodal search and domain-specific embeddings. - Secure evaluation frameworks that become organizational standards. Compensation and Application Process: Digital Green is a great place to work and prides itself on a competitive and comprehensive compensation and benefits package. Digital Green Trust is proud to be an equal opportunity employer. All qualified applicants will receive consideration for employment without regard to race, color, religion, sex, national origin, disability status, protected veteran status, or any other characteristic protected by law. Please send your application along with an updated CV and a cover letter explaining your suitability for the position. All completed applications should be applied through Join us at Digital Green and leverage your expertise in AI to transform agricultural practices and improve the livelihoods of smallholder farmers worldwide.