Kaleem Ullah Qasim — AI Researcher

I build agentic AI systems that work in production and publish research on why they work. 13+ publications, 33 citations, first-author work in JAIR. Most AI projects fail at implementation. I bridge the gap between cutting-edge research and deployed systems: agentic RAG pipelines processing 10K+ documents with graph-based retrieval, multi-agent orchestration workflows that actually complete tasks, and NL2SQL interfaces that turned 5-minute manual queries into 10-second automated ones. Currently finishing my PhD at Southwest Jiaotong University, focused on LLM reasoning, reinforcement learning with verifiable rewards, and multi-agent systems. My published frameworks RDoLT (recursive decomposition, JAIR), CARD (complexity-aware reasoning), MARBLE (multi-agent rule-based reasoning)directly inform the production systems I build. 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 → 13+ publications | 33 citations | Venues: JAIR, IEEE INFOCOM, Alexandria Engineering Journal → Focus areas: LLM reasoning, context engineering, multi-agent orchestration, RLHF & post-training alignment → First-author: RDoLT (recursive decomposition), CARD (complexity-agnostic reasoning), MARBLE (multi-agent engine), VERIFY-RL (verifiable mathematical reasoning) → Named systems: TraffiCoT-R (spatio-temporal CoT), CORTEX-V (cognitive reasoning), LLMFacility (evolutionary optimization) 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗦𝘆𝘀𝘁𝗲𝗺𝘀 → Agentic RAG & document Q&A with 92% accuracy using hybrid retrieval and knowledge graphs → Multi-agent orchestration with stateful memory: 87% human-quality scores, 65% faster production → SQL chat agents: 79% faster response times through context-aware query optimization → LLM alignment & post-training: DPO, GRPO, DAPO, RLVR pipelines for enterprise deployments → Fine-tuning (LoRA/QLoRA/SFT) with GDPR-compliant local LLM deployment 𝗧𝗲𝗰𝗵 𝗦𝘁𝗮𝗰𝗸 LangChain • LangGraph • CrewAI • AutoGen • DSPy • MCP • A2A • Pinecone • Weaviate • GPT-4 • Claude • DeepSeek • Qwen • Llama • Python • FastAPI • PyTorch • Kubernetes • AWS • Azure ML On Upwork: Top Rated, 100% job success, every review 5 stars. Open to: research collaborations, AI consulting, and interesting problems at the intersection of agentic AI research and production systems. Languages: English (fluent) • Chinese (HSK5) • Urdu (native) • Hindi (fluent)

Stackforce AI infers this person is a skilled AI researcher and engineer specializing in cybersecurity and multi-agent systems.

Location: Chengdu, Sichuan, China

Experience: 1 yr 6 mos