Machine Learning & Generative AI Engineer

Job Description

Build intelligent AI systems, production-ready LLM applications, agentic workflows, and scalable machine learning solutions that transform ideas into real-world impact. You will design, develop, and deploy machine learning models, LLM-powered applications, retrieval-augmented generation (RAG) pipelines, and autonomous AI agents. The role spans across the entire AI stack—from data processing and model experimentation to production deployment, evaluation, and optimization. You will collaborate closely with product designers, Flutter developers, backend engineers, and product teams to deliver impactful AI experiences while keeping up to date with emerging research and open-source models.

* Foundation Models and Fine-Tuning (Open-Source)
* Open-Source LLMs: Llama (Meta), Mistral / Mixtral, Qwen, Gemma, Phi.
* Fine-Tuning and Adaptation: PEFT (Parameter-Efficient Fine-Tuning), LoRA / QLoRA, and model quantization techniques (e.g., GPTQ, AWQ, GGUF) to run and adapt models efficiently.
* Multimodal and Specialized Models: Vision-Language Models (VLMs), speech recognition, and text-to-speech open architectures.
* LLM Inference and Serving Infrastructure
* Local and High-Performance Inference Engines: Ollama, vLLM, and llama.cpp for optimized, low-latency model serving.
* GPU Computing and Optimization: CUDA, PyTorch/TensorFlow optimizations, distributed inference, and model serving runtimes.
* RAG Pipelines and Agentic Workflows (Orchestration)
* Orchestration Frameworks: LangChain, LlamaIndex, or custom Python-based agent architectures.
* Autonomous Agents: Frameworks for building multi-agent systems, tool-use reasoning loops (e.g., ReAct), and autonomous decision-making pipelines.
* Vector Databases and Retrieval (Open-Source): Chroma, FAISS, Milvus, Weaviate, or PGVector (PostgreSQL extension) for semantic search and retrieval-augmented generation.
* Experimentation, Observability and MLOps
* Tracking and Experimentation: MLflow or Weights and Biases (W&B) for tracking metrics, hyperparameters, and model runs.
* LLM Evaluation and Observability: LangSmith or open-source equivalents for tracing agent steps, debugging prompts, and evaluating RAG accuracy.
* Core Engineering and Backend Stack
* Programming Language: Python (primary language for data processing, model experimentation, and AI pipelines).
* Backend Frameworks: FastAPI or Flask for exposing model endpoints and microservices.
* Collaboration and Delivery: Git, CI/CD pipelines, and automated testing frameworks to ship robust, production-ready code alongside Flutter and backend engineering teams.