Job Description
Description : Were Hiring ANN / Vector Database Engineer.
Location : Remote (India).
Company : HYI.AI.
Experience : 4+ Years.
Notice Period : Immediate Joiners Preferred.
Key Role Description :
- Extend our database with native vector capabilities including data types, ANN indexes, query planning, and developer APIs.
- Enable applications to run hybrid (sparse + dense) search at production scale with ACID guarantees.
Key Responsibilities :
- Implement vector data model features : VECTOR/EMBEDDING column types, distance operators (L2, cosine, dot), and server-side normalization.
- Build ingestion/backfill pipelines for embeddings, including bulk loaders and online updates.
- Develop production-grade ANN indexing (HNSW / IVF-Flat / IVF-PQ) in C/C++, integrated with WAL, checkpoints, and snapshots for crash safety and fast recovery.
- Manage index lifecycle : online CREATE INDEX, background build/merge, partial rebuilds, compaction/vacuum, tombstone handling.
- Ensure transactional consistency : MVCC, visibility rules, point-in-time queries, and predictable latency under mixed workloads.
- Integrate query planner/optimizer : new planner nodes, hybrid plans, predicate pushdown, join strategies, and parameterized search.
- Support distributed operations : replication, shard placement/rebalancing, hot-index warmup, and tenant-level resource isolation.
- Build APIs, SDKs, and admin tools : SQL syntax, gRPC/HTTP endpoints, Python/Java/Node SDKs, collection/index management, backups, zero-downtime upgrades.
- Ensure observability : metrics, traces, slow-query analysis, pg_stat-like views for ANN workloads.
Minimum Qualifications :
- 4+ years building performance-critical systems in C/C++, plus experience in Python, Node.js, or Java.
- Strong knowledge of data structures, concurrency, memory layout, profiling, and low-latency I/O.
- Hands-on experience with search/vector stacks such as FAISS, hnswlib, ScaNN, NMSLIB, Milvus, Qdrant, Vespa, Lucene/Elasticsearch/OpenSearch.
- Experience integrating with database internals : WAL, MVCC/ACID, query planning, storage engines.
Nice to Have :
- SIMD (AVX2/AVX-512/NEON), GPU/CUDA/ROCm, or FPGA acceleration.
- Knowledge of LSM/B-tree/columnar storage, compaction strategies, snapshot/backup/restore.
- Experience with re-ranking (MMR, late interaction/ColBERT), PQ/OPQ/SQ quantization, workload-aware autotuning.
- Multi-tenant resource isolation, k8s operations, packaging (static, musl, cross-compile).
- Developer tooling craftsmanship (docs, CLIs, codegen, examples).
- Exposure to generative AI is an advantage.
Why Join HYI.AI :
- Work on cutting-edge vector search and AI infrastructure.
- Join a collaborative remote-first work culture.
- Opportunity to build enterprise-grade vector search engines used at production scale.
Did you find something suspicious?
Posted By
4
Posted in
Others
Functional Area
Database Admin / Development
Job Code
1559486
