About the Role:

We are looking for a hands-on AI/ML Engineer to design, build, and deploy multilingual NLP and speech systems, including Neural Machine Translation (NMT), Speech-to-Text (STT), and Text-to-Speech (TTS). This role combines applied research and engineering with a robust focus on multilingual modeling, low-resource learning, and offline/on-device (edge) deployment.

Key Responsibilities:

Multilingual Modeling &

- NLP Systems:
- Design and develop Machine Translation (NMT), Speech-to-Text (STT), and Text-to-Speech (TTS) systems.
- Work with transformer-based and sequence-to-sequence architectures.
- Adapt models for multilingual and low-resource scenarios.

Data Engineering &

- Processing:
- Build pipelines for data ingestion, cleaning, and preprocessing.
- Handle low-resource and noisy datasets.
- Implement data augmentation, back-translation, and synthetic data generation.
- Work with JSON, Parquet, TFRecords.

Model Training &

- Optimization:
- Train models using PyTorch, TensorFlow, or JAX.
- Use Hugging Face Transformers, Fairseq, OpenNMT.
- Apply transfer learning and PEFT (LoRA, Adapters).
- Optimize with mixed precision and distributed training.
- Improve latency, throughput, and memory efficiency.

Evaluation &

- Experimentation:
- Design evaluation pipelines (BLEU, WER, ROUGE, METEOR).
- Use MLflow, Weights &
- Biases, TensorBoard.
- Perform benchmarking, error analysis,



and regression testing.

Deployment &

- Edge Systems:
- Deploy models in offline/on-device environments.
- Build inference APIs using FastAPI, Flask, or gRPC.
- Work with Docker and Kubernetes.
- Optimize inference using ONNX, TensorRT, TorchScript.

AI Systems &

- MLOps:
- Develop pipelines using Airflow or Prefect.
- Implement CI/CD (GitHub Actions, Jenkins).
- Monitor model performance and drift.

Research &

- Innovation:
- Explore multilingual LLMs and speech-language models.
- Prototype and evaluate new approaches.
- Contribute to internal frameworks and tools.

Collaboration:

- Work with ML engineers, data scientists, and research teams.
- Collaborate with linguistic and annotation teams.
- Document systems and experiments.

Tech Stack &

- Good to Have:
- Advanced Concepts: Multilingual LLMs, Transfer learning, domain adaptation, Real-time inference systems.
- Additional Tools: FAISS, Milvus, Weaviate, Airflow, Prefect, GitHub Actions, Jenkins, DeepSpeed, FSDP.

What We're Looking For:

- Strong problem-solving skills.
- Ability to work in research-driven environments.
- Hands-on builder mindset.
- Good communication skills.

Why Join Us:

- Work on cutting-edge multilingual AI and speech systems.
- Build production-grade AI solutions.
- Exposure to LLMs, speech AI, and edge deployment.

📌 AI/ML Engineer - NLP & Speech Recognition (India)
🏢 MCLAREN STRATEGIC VENTURES INDIA PRIVATE
📍 India