0
Jobs currently available
0
Portfolio companies
Senior DSP & Bid-Optimization Engineer (Rust/Scala)
Teza
Software Engineering
Posted on Dec 11, 2025
👩💻
Senior DSP & Bid-Optimization Engineer (Rust/Scala)
About LoudEcho
LoudEcho is an AI ad platform/DSP that buys media, autonomously generates creative, and optimizes every impression in real time—sitting on the bidding stream to analyze live context and audience signals.
Our platform supports AI-generated creatives and uses contextual signals, consent frameworks, and privacy-compliant data to select and serve ads.
Safety is by design, with multi-layer brand-safety controls and brand-defined constraints before an ad is shown.
📊 What you’ll spend your time on
Real-time bidding (RTB) – author low-latency Rust services that process 100 k+ QPS, manage multi-imp auctions, and integrate exchange adapters (ORTB 2.5 & 2.6).
Campaign selection & pacing – design online allocators, budget throttling, and frequency capping that react within 5 ms.
Offline optimization – build Spark/ClickHouse jobs that crunch billions of impressions to feed TorchScript/TensorFlow Serving models.
Model inference at scale – embed gRPC/REST inference calls (TorchScript, TFS) into the hot path with sub-millisecond budget.
LLM acceleration (bonus) – prototype GPT-style creative scoring / keyword expansion pipelines that run in-process or via vLLM-based micro-services.
End-to-end ownership – from spec → PR → prod rollout in days; measure impact with A/B tests and Grafana.
👩💻 Capabilities
7 + years engineering B2B ad-tech (DSP, bidder, SSP, ad server) and ownership of at least one high-throughput exchange integration.
Expert-level Rust (or Scala/C++/Go with evidence of Rust ramp-up). You know how to squeeze every μs with lock-free queues, SIMD, and memory pools.
Build and scale real-time feature stores / enrichment layers (Aerospike, Redis, Scylla or similar) that serve per-user state with sub-millisecond P99.
Hands-on shipping of streaming + columnar stacks (Kafka/Pulsar, Flink/FastAPI, ClickHouse/BigQuery).
Cloud Ops & FinOps - Running Kubernetes workloads on AWS or GCP and cutting costs via Kubecost, Spot.io, AWS Cost Explorer, etc.
Proven design of online + offline optimisation loops (Multi-armed bandit, Gradient-based ROAS optimisation, Kalman pacing, etc.).
Solid ML inference chops – deploying TorchScript or TFS models behind gRPC/REST with autoscaling.
Comfortable writing Vibe level code (clarity & tests first, but “ship Friday” is a feature).
Startup mindset – choose pragmatism over purity, communicate status & trade-offs plainly.
Nice-to-Haves
Real-time LLM integration (vLLM, GGUF, Triton).
GPU scheduling / CUDA graph experience for inference surge control.
How We Work
Shipping cadence: idea → spec review (24 h) → PR (48 h) → staged rollout behind feature-flag.
Tooling: GitHub + Actions, Bazel, k8s, Grafana/Loki.
Competitive salary + meaningful equity.
🏓 Contact
Excited to join us? send your CV here: careers@teza.ai