Solytics Partners - Senior Data Scientist - GenAI

About Us :

Solytics Partners is a Global Analytics firm, recognized with multiple industry awards for innovation and excellence. Our team comprises experts with deep domain knowledge in risk, analytics, AI/ML, AML/FCC, and fraud. By converging this expertise with cutting-edge technologies like AI, Machine Learning, Generative AI, and Large Language Models (LLMs), we deliver powerful automated platforms and incisive point solutions.

Our offerings enable clients to streamline and future-proof their risk, AML, and analytics processes, comply seamlessly with global regulations, and safeguard financial systems. Whether it's solving complex challenges or driving operational efficiency, Solytics Partners is committed to empowering organizations with transformative tools to stay ahead in an evolving regulatory landscape.

About the role :

We are seeking a highly capable GenAI Data Science Engineer with strong foundations in Data Science, Machine Learning, Natural Language Processing, and Deep Learning, combined with hands-on expertise in LLM engineering and production software development.

The ideal candidate must be able to independently design, architect, develop, optimize, deploy, and maintain production-grade AI/ML and Generative AI solutions. This role requires a combination of analytical depth, strong Python engineering, API development, and practical understanding of LLM behavior, tokenization, inference, evaluation, and performance optimization.

We are looking for an engineer who can go beyond prototypes and framework integrations to understand the underlying technology, make sound architectural decisions, solve complex technical problems, and deliver reliable, measurable business outcomes.

Key responsibilities :

- Design and implement end-to-end AI/ML and Generative AI solutions, from problem formulation and data preparation through deployment and production monitoring.

- Build LLM-powered applications using commercial APIs and/or open-source models, selecting the appropriate approach based on quality, latency, cost, and operational requirements.

- Develop production-ready REST APIs and backend services using Python, FastAPI, or Flask, with appropriate validation, authentication, asynchronous execution, error handling, and testing.

- Apply strong ML, NLP, and Deep Learning fundamentals to model selection, experimentation, feature engineering, evaluation, and optimization.

- Implement LLM workflows involving prompt engineering, tokenization, embeddings, structured outputs, function calling, and context-window management.

- Design and optimize RAG pipelines, including document processing, chunking, retrieval, hybrid search, reranking, and grounded response generation.

- Optimize token consumption, context utilization, latency, throughput, and API/inference costs while maintaining acceptable output quality.

- Establish systematic evaluation and regression testing for model quality, retrieval relevance, groundedness, hallucinations, and application reliability.

- Implement robust production practices, including logging, monitoring, retries, timeouts, caching, versioning, and failure recovery.

- Independently troubleshoot complex issues and own the complete solution lifecycle, including architecture, development, deployment, and continuous improvement.

Mandatory technical requirements :

Python and software engineering :

- Strong hands-on Python programming, including object-oriented programming, data structures, algorithms, modular design, testing, and debugging.

- Ability to write clean, maintainable, reusable, and production-quality code.

- Experience integrating external services and building reliable backend applications.

Data Science, ML, NLP and Deep Learning :

- Strong understanding of statistics, probability, exploratory data analysis, feature engineering, and experimental design.

- Practical knowledge of supervised and unsupervised learning, model selection, validation, evaluation metrics, and error analysis.

- Solid understanding of neural networks, transformers, attention mechanisms, NLP, embeddings, and semantic similarity.

- Hands-on experience with PyTorch or TensorFlow, or an equivalent deep learning framework.

Generative AI and LLM engineering :

- Demonstrable experience developing LLM-powered applications beyond basic prompt experimentation.

- Strong conceptual understanding of tokenization, context windows, embeddings, inference behavior, model limitations, and model selection.

- Practical experience with prompt design, structured generation, tool/function calling, and robust LLM API integrations.

- Ability to evaluate LLM outputs systematically and diagnose failures involving retrieval, generation, grounding, and formatting.

API development and optimization :

- Strong REST API development experience using FastAPI or Flask.

- Ability to implement request validation, asynchronous processing, exception handling, retries, timeouts, and automated tests.

- Demonstrable understanding of token usage optimization, context management, latency reduction, throughput, and cost-performance trade-offs.

- Ability to benchmark performance, establish baselines, and validate optimization improvements.

Production deployment and ownership :

- Proven hands-on experience deploying and maintaining AI/ML or GenAI applications in a production environment.

- Understanding of application monitoring, logging, troubleshooting, reliability, and release management.

- Ability to independently take ownership of a solution from technical design through production support.

- Hugging Face Transformers and modern open-source LLM ecosystems.

- RAG frameworks such as LangChain or LlamaIndex.

- Vector databases such as FAISS, Chroma, Pinecone, or pgvector.

- Fine-tuning and parameter-efficient techniques such as LoRA/QLoRA.

- Docker, cloud platforms, CI/CD, MLOps, and observability tools.

- Model-serving and inference optimization technologies such as vLLM, where relevant.

What success looks like :

The successful candidate can independently deliver a complete AI solution, justify model and architecture choices, demonstrate measurable improvements in quality or performance, and maintain the solution reliably in production.