Machine Learning Engineer Full Time NEW

Company Description

Think Right is one of the leading provider of cutting-edge technology solutions designed to help businesses enhance scalability, security, and operational efficiency. We specialize in advanced data tools, custom application development, security solutions, and end-to-end data management. Backed by a team of experts in software developers, data engineering, artificial intelligence, research, and data science, we pride ourselves on delivering tailored solutions. Join Think Right to be a part of a forward-thinking organization driving advancements in data technology.

Role Description

We are seeking a full-time Machine Learning (ML) Engineer to join our team. This hybrid role is based in Bengaluru, with flexibility for partial remote work. The ML Engineer will be responsible for designing and implementing machine learning models, optimizing algorithms, conducting pattern recognition analyses, and leveraging neural networks for data-driven solutions. The ML Engineer owns the model layer – from speech-to-text accuracy to fine-tuning base LLMs on clinical conversation data. You will work closely with the AI Architect and AI Engineer to build models that meet the accuracy bar required for clinical documentation. This means rigorous experimentation, robust pipelines, and a deep understanding of what clinical correctness actually means.

Key Responsibilities

• Own the speech-to-text pipeline – evaluate, adapt, and optimise models (Whisper, AWS Transcribe Medical) for psychiatric and behavioural health vocabulary
• Build and manage speaker diarization pipelines to correctly attribute therapist vs patient speech in session recordings
• Design and execute fine-tuning workflows for clinical summarisation and note generation tasks – using LoRA/QLoRA on open-source LLMs (Llama, Mistral)
• Develop training data pipelines: collection, anonymisation (de-identification of PHI), annotation tooling, and quality review
• Build MLOps infrastructure – experiment tracking (MLflow/W&B), model registry, automated evaluation, and deployment pipelines
• Establish and maintain model performance benchmarks: transcription WER, SOAP note accuracy, hallucination rate, and clinical relevance scores
• Implement drift detection and monitoring for production models – surface degradation early before it impacts clinical outputs
• Research and evaluate new model releases and architectures relevant to the AISpire360 use case
• Collaborate with AI Engineer on embedding strategies, retrieval quality, and semantic search performance
• Ensure all ML systems meet HIPAA requirements for data handling, storage, and audit logging

Qualifications

• 5+ years in machine learning engineering with production model deployment experience
• Strong PyTorch skills – training, fine-tuning, and inference optimisation
• understanding of Statistics and its application in data modeling.
• Experience with speech-to-text systems: Whisper, wav2vec 2.0, or cloud equivalents (AWS Transcribe Medical, Azure Speech Services)
• Hands-on experience fine-tuning LLMs using PEFT techniques (LoRA, QLoRA, instruction tuning)
• Proficiency in Hugging Face ecosystem – Transformers, PEFT, Datasets, Evaluate
• Strong knowledge of MLOps: MLflow, Weights & Biases, DVC, or equivalent
• Experience building data pipelines for NLP tasks – tokenisation, annotation, augmentation, and quality control
• Solid understanding of evaluation methodology for generative models (ROUGE, BERTScore, custom clinical evals)
• Knowledge of model optimisation for production: quantisation, distillation, and batched inference
• Strong expertise in Pattern Recognition and Neural Networks.

Nice to Have

• Experience with clinical NLP or medical language models (BioBERT, ClinicalBERT, Med-PaLM)
• Background in de-identification and PHI scrubbing pipelines for healthcare data
• Familiarity with HIPAA and HITRUST compliance requirements for ML systems
• Experience deploying models on AWS SageMaker, GCP Vertex AI, or equivalent managed platforms