AI Engineer - 2

• SwiftChat (the platform on which we build with ecosystem partners)
• SwiftPAL - validated by a two-year RCT led by Nobel laureate Prof. Michael Kremer (Development Innovation Lab, University of Chicago), whose team found students gained "almost an additional 2 years of extra learning" in just 17 months
• Swiftee (the AI learning tutor built for Bharat).

Key Responsibilities

• Design, develop, and deploy production-grade AI/ML systems using LLMs, NLP, Computer Vision, Speech, and multimodal models.
• Build agentic AI applications and workflows using tool calling, planning, memory, retrieval, state management, and human-in-the-loop mechanisms.
• Develop RAG and knowledge-grounded systems using embeddings, retrieval, reranking, and context optimization.
• Experiment with model adaptation techniques such as SFT, LoRA/QLoRA, synthetic data generation, prompt optimization, and knowledge distillation.
• Design evaluation pipelines for accuracy, groundedness, hallucination, safety, latency, reliability, and cost.
• Integrate AI systems with APIs, databases, vector stores, and backend services.
• Collaborate with product, engineering, research, and education teams to take AI solutions from prototype to production.

Qualifications

• Bachelor's or master's degree in Computer Science, Engineering, AI/ML, or a related field.
• 2–4 years of industry experience building machine learning or AI systems.
• Strong understanding of machine learning, deep learning, transformers, embeddings, and LLMs.
• Strong programming skills in Python with hands-on experience using PyTorch or similar frameworks.
• Experience building real-world applications in one or more areas such as LLMs, RAG, NLP, Computer Vision, Speech, OCR, or multimodal AI.
• Experience with LLM fine-tuning or model adaptation techniques such as SFT, LoRA/QLoRA, distillation, or preference optimization is preferred.
• Exposure to agentic AI systems, including tool calling, workflow orchestration, memory, planning, or multi-agent systems.
• Familiarity with frameworks such as LangGraph, LangChain, or similar.
• Experience with model evaluation and defining meaningful offline and production metrics.
• Familiarity with Docker, cloud platforms, APIs, CI/CD, model serving, and monitoring is preferred.
• Strong problem-solving skills and ability to independently experiment, debug, and improve AI systems.

• Own AI features end-to-end — research through deployment and iteration.
• Write clean, tested, production-quality code.
• Prototype fast and decide based on measurable results.
• Balance quality, reliability, latency, cost, and safety — not just demos.
• Stay current with LLM, agentic, and multimodal AI developments.
• Communicate technical trade-offs clearly across teams.
• Own the reliability and safety of systems deployed at scale, especially in child-facing educational contexts.

• Work with a diverse, passionate team spread across India.
• Gain exposure to large-scale, real-world impact through our presence in 2/3rds of India.
• Contribute directly to learning outcomes for millions of children.
• Build future-focused skills at the intersection of education, technology, and public systems.
• Thrive in a culture of ownership, learning, and collaboration.
• Enjoy stability with health insurance and other employee-friendly benefits.