kAIgentic
kAIgentic
kAIgentic is a Singapore-headquartered enterprise AI SaaS company helping complex, knowledge-intensive organizations turn tacit operational knowledge into governed AI agents. Its platform combines agent configuration, diagnostic analysis of people, systems, and processes, and a runtime environment for deployment with human oversight, telemetry, and ongoing refinement. With roots in Japan and enterprise financial-services collaboration, kAIgentic focuses on assurance, compliance, and practical AI modernization for regulated businesses without requiring a rip-and-replace approach.

Senior AI Research Software Engineer – kAIgentic (Hybrid, India)

Lead the development of governed knowledge graphs, learning systems, and production AI pipelines at kAIgentic. Build reliable, evidence-grounded solutions for regulated enterprise operations.

Description

  • Lead the full design and implementation of a knowledge-graph schema covering hierarchy, provenance, versioning, and extensibility.
  • Develop grounded-extraction pipelines that connect every claim to source evidence and an epistemic status.
  • Create and refine multi-context inference harnesses for memory management, budget control, and large-scale compaction.
  • Build knowledge-distillation workflows that bring external data into the internal ontology.
  • Develop consolidation frameworks that unify documents, interviews, and event logs while resolving conflicts.
  • Create and refine model-adaptation workflows using SFT, DPO, and RFT.
  • Develop learning-loop infrastructure that captures feedback, supports active learning, and updates the ontology continuously.
  • Build reinforcement-learning environments and curricula with reward signals and evaluation measures for safe operation.
  • Develop production retrieval-generation systems within defined quality, latency, and cost targets.
  • Create evaluation frameworks using blind rubrics and LLM-as-judge validation to assess grounding and hallucination.

Requirements

  • Demonstrated success taking AI research systems from initial prototype through production in regulated environments.
  • Work with an AI-native approach and high execution velocity.
  • Advanced Python and PyTorch expertise, including large-scale fine-tuning with SFT, DPO, and RFT.
  • Expertise in graph databases and ontology engineering, including schema architecture, version control, and provenance.
  • Practical experience building extraction systems that associate claims with source evidence and epistemic status.
  • Hands-on experience with reinforcement-learning environments, curriculum development, and reward modeling for autonomous systems.
  • Strong command of retrieval-augmented generation, latency reduction, and cost-efficient scaling.
  • Ability to conduct rigorous applied research through hypothesis design, ablation testing, and statistical validation.
  • Sound judgment when choosing between model adaptation and context engineering, with strong cross-functional communication.
  • PhD in Artificial Intelligence, Computer Science, or a related discipline.
  • At least four years of relevant industry research experience.

Benefits

  • Collaborate globally with colleagues across Singapore, India, Japan, Europe, and the United States.
  • Learn and develop alongside experienced leaders from major enterprises.
  • Take ownership of meaningful work from the first day.
  • Work in an environment that values psychological safety, candid debate, and disciplined experimentation.
  • Combine startup speed with work on enterprise-scale infrastructure.
  • Help influence the product, company culture, and customer results.
  • Address mission-critical challenges in regulated enterprise environments.

Related Jobs

The Duke of Edinburgh's Award

Business Support Assistant, Events and Ambassadors

The Duke of Edinburgh's Award
201 – 500 Employees
CharityEducation

Provide event administration, coordinate suppliers and travel, and respond to enquiries for The Duke of Edinburgh’s Award. This hybrid UK role is home based, with weekly attendance at the Reading office.

Open