AI Research Engineer

A fast paced startup Company is hiring an AI Research Engineer.

Job Type: Full-time, onsite

Shift Timings: 3 pm to 12 am

Location: Shahrah-e-Faisal, Karachi

Industry: AI / Fintech

Responsibilities

· Research and evaluate techniques for improving LLM reasoning, accuracy, and efficiency — including prompting strategies, fine-tuning, distillation, and model selection

· Design and run rigorous experiments with proper baselines, ablations, and statistical analysis to measure the impact of each technique

· Build evaluation frameworks and benchmarks to systematically measure model quality across Yara's key tasks

· Improve retrieval-augmented generation pipelines — embedding models, chunking strategies, re-ranking, query decomposition, and context window optimization

· Explore fine-tuning approaches (LoRA, RLHF, DPO) to adapt foundation models for domain-specific performance

· Study and reduce failure modes — hallucination, instruction drift, reasoning errors, and inconsistent structured output generation

· Stay current with the research landscape (papers, open-source releases, model updates) and identify techniques with high applicability to Yara's use cases

· Prototype emerging approaches (chain-of-thought, self-consistency, tool-use planning, multi-agent coordination) and measure real-world impact

· Translate research findings into production-ready improvements and document results for the engineering team

Requirements:

· 4-5 years in ML/AI research or applied AI engineering

· Strong understanding of LLM architectures, training dynamics, and inference optimization

· Hands-on experience with fine-tuning (LoRA, RLHF, DPO) and systematic model evaluation

· Proficient in Python and ML frameworks (PyTorch,HuggingFace Transformers)

· Experience designing experiments with proper controls, ablations, and reproducibility

· Familiarity with RAG systems, embedding models, and retrieval pipelines

· Strong written communication — ability to produce clear internal research reports and technical documentation

Interested candidates may apply.

Application Question(s):

• Are you willing to work onsite from 3 pm to 12 am ?

Education:

• Bachelor's (Preferred)

Experience:

• ML/AI research: 4 years (Preferred)
• ML frameworks: 4 years (Preferred)
• fine tuning and systematic model evaluation: 4 years (Preferred)
• RAG system: 2 years (Preferred)

Work Location: In person