ShyftLabs
Website:
shyftlabs.io
Job details:
ShyftLabs is a growing data product company that was founded in early 2020 and works primarily with Fortune 500 companies. We deliver digital solutions built to help accelerate the growth of businesses in various industries, by focusing on creating value through innovation.
Job Responsibilties:
- 4–8 years of relevant experience in LLMs, Backend Engineering, and MLOps.
- LLM Expertise
- Model Fine-tuning: Experience with parameter-efficient fine-tuning methods (LoRA, QLoRA, adapter layers)
- Inference Optimization: Knowledge of quantization, pruning, caching strategies, and serving optimizations
- Prompt Engineering: Prompt design, few-shot learning, chain-of-thought prompting, and retrieval-augmented generation (RAG)
- Model Evaluation: Experience with AI evaluation frameworks and metrics for different use cases
- Monitoring & Testing: Design of automated evaluation pipelines, A/B testing for models, and continuous monitoring systems
- Backend Engineering
- Languages: Proficiency in Python, with experience in FastAPI, Flask, or similar frameworks
- APIs: Design and implementation of RESTful APIs and real-time systems
- Databases: Experience with vector databases and traditional databases
- Cloud Platforms: AWS, GCP, or Azure with focus on ML services
- MLOps & Infrastructure
- Deployment: Experience with model serving frameworks (vLLM, SGLang, TensorRT)
- Containerization: Docker and Kubernetes for ML workloads
- Monitoring: ML model monitoring, performance tracking, and alerting systems
- Evaluation Systems: Building automated evaluation pipelines with custom metrics and benchmarks
- CI/CD: MLOps pipelines for automated testing, and deployment
- Orchestration: Experience with workflow tools like Airflow.
Preferred Qualifications:- LLM Frameworks: Hands-on experience with Transformers, LangChain, LlamaIndex, or similar
- Monitoring Platforms: Knowledge of LLM-specific monitoring tools and general ML monitoring
- Distributed Training and Inference: Experience with multi-GPU and distributed training and inference setups
- Model Compression: Knowledge of techniques like distillation, quantization, and efficient architectures
- Production Scale: Experience deploying models handling high-throughput, low-latency requirements
- Research Background: Familiarity with recent LLM research and ability to implement novel techniques
- Tools & Technologies We Use
- Frameworks: PyTorch, Transformers, TensorFlow
- Serving: vLLM, TensorRT-LLM, SGlang, OpenAI API,
- Infrastructure: Kubernetes, Docker, AWS/GCP
- Databases: PostgreSQL, Redis, Vector DBs
We are proud to offer a competitive salary alongside a strong insurance package. We pride ourselves on the growth of our employees, offering extensive learning and development resources.
Click on Apply to know more.