Recommendation Systems Engineer
Experience: 3.0-8 years Location: BangaloreCTC: 30 –60L
Senior ML/AI Engineer with 3-8 years experience in recommendation systems and ranking models.
About this role
About the Role
Dashreels streams bite-sized, emotionally addictive AI-generated dramas — 70-minute series told through 1–2 minute episodes. New episodes can be spun up hourly by creators and algorithms alike, making discovery as much an ML problem as a storytelling one.
We're looking for a RecSys specialist to own the ranking and discovery engine that surfaces the right episode for every viewer, moment by moment. This is not a full-stack role — we want someone who lives and breathes recommendation systems, ranking models, and real-time ML.
What You'll Do
- Design, build, and iterate production ranking and recommendation systems — candidate retrieval, feature stores, real-time scoring, and model serving
- Invent Gen-AI-native ranking techniques — leverage multimodal embeddings, LLM-generated summaries, and agentic diversifiers to rank freshly generated episodes without historical signals
- Build and optimize data pipelines processing billions of engagement events (watch-time, completion, scroll depth)
- Own experimentation — offline evaluation harnesses, weekly A/B tests, causal ML for measuring ranking impact
- Tune systems for sub-50ms latency at scale while keeping infra costs predictable
- Own viewer metrics: watch-time, completion rate, discovery depth, content diversity
What We're Looking For
- 3–8 years engineering experience with at least one recommender or ranking system in production
- Strong Python skills; familiarity with Java/Kotlin for serving is a plus
- Experience with ranking models — multi-task learning, GBDT, deep ranking, multi-armed bandits
- Hands-on with real-time data pipelines (Spark, Kafka, Flink, or equivalent)
- A/B testing platforms and understanding of causal ML
- Experience with feature stores, embeddings, and candidate retrieval at scale
Bonus
- Experience with LLM-powered retrieval or agent-style recommenders (LangGraph, AutoGen, ReAct)
- Integrating multimodal embeddings (diffusion, TTS, image models) into ranking pipelines
- Background in content/entertainment platforms (short video, OTT, social feeds)
- Experience at companies like ShareChat, Moj, Josh, Roposo, Dailyhunt, or similar content platforms