AI Product Manager KAI (KGeN)
Weblink: https://kgen.io/, Kai: https://kai.kgen.io/
Location: Bangalore, India
Team: KAI Training & Evaluation
Role Overview
KGeN is building a large-scale multimodal AI annotation and evaluation platform (KAI).
We are hiring an AI Product Manager who understands labeling workflows, model-assisted annotation, and applied AI systems. You will work closely with engineers, ML researchers, and annotator operations to design and ship features that improve annotation accuracy, speed, and cost efficiency.
This is an applied AI product role ideal for someone who has worked with annotation tools and ML-powered features, and can translate AI capabilities into scalable product workflows.
Key Responsibilities
AI-Assisted Annotation
- Define product features for pre-annotation using models such as WhisperX, SAM/SAM2, CLIP, GroundingDINO, Qwen2-VL, and LLaVA.
- Prioritize and deliver model-assisted labeling workflows (auto-bbox, transcription, tagging, segmentation).
- Work with ML and engineering teams to integrate AI capabilities into the annotation platform.
Annotation Workflow & Platform Design
- Own task flows, worker assignment logic, reviewer flows, QC, escalation, and consensus scoring.
- Design and optimise skill-based routing and workforce management systems.
- Improve overall task throughput, quality, and operator experience.
Model Evaluation & Quality Systems
- Define dashboards, leaderboards, and reporting for model performance.
- Work with ML Research Engineers to establish metrics like IoU, mAP, WER, CER, DER, BLEU, and ROUGE.
- Use model evaluation insights to shape feature decisions and fallback logic.
Technical Product Requirements
- Write detailed PRDs, technical specs, workflows, and data schemas.
- Manage backlog, sprints, and cross-functional delivery with engineering.
- Define KPIs for quality, turnaround time, model performance, and cost per task.
Cross-Functional Collaboration
- Collaborate with engineering, ML research, operations, and enterprise stakeholders.
- Communicate complex model behavior and tradeoffs in simple terms.
- Translate customer requirements into feature definitions for the platform.
Data-Driven Decisions
- Use SQL/BigQuery/Athena and dashboards to analyze quality, throughput, and performance.
- Identify bottlenecks and opportunities for automation or ML enhancement.
- Validate feature impact using metrics-based analysis.
Required Qualifications
- 35 years of product management experience, with at least 12 years in AI/ML-driven products.
- Strong understanding of core ML concepts (classification, detection, embeddings, multimodal models, model evaluation).
- Hands-on experience with annotation tools such as:
- Label Studio
- SuperAnnotate
- Labelbox
- Encord
- CVAT
- Experience defining and shipping workflow-heavy or operations-heavy products.
- Ability to write strong PRDs and work directly with engineering teams.
- Working knowledge of SQL and data analysis tools.
- Experience working with ML engineers or applied research teams.
Preferred Experience
- Managed or built workflows for annotation, quality control, or HITL pipelines.
- Experience designing consensus systems, reviewer pipelines, or routing logic.
- Familiarity with HuggingFace, OpenAI, or cloud-based ML platforms.
- Exposure to multimodal data (images, audio, text, video).
- Previous startup or fast-moving tech environment experience.
What Success Looks Like
- Measurable reduction in annotation cost through model-assisted workflows.
- Improved accuracy and throughput for annotation teams.
- Clear, reliable dashboards for model and annotator performance.
- Smooth end-to-end workflows supporting thousands of annotators.
- Strong coordination between engineering, ML research, and operations.
- Timely delivery of high-impact AI features aligned with KAI's roadmap.
