21 Nov
|
PinnacleU
|
Bengaluru
21 Nov
PinnacleU
Bengaluru
Position: AI Engineer
Location: Bangalore
Working Days: Monday to Friday
Experience Required: 3-5 years
Our Client is building an AI-native Social Content & Campaign Intelligence OS for global consumer brands , helping them monitor, audit, reason over, and activate intelligence across social content, influencer campaigns, UGC, and commerce. We are working with Fortune 500 global enterprise brands.
The role
You'd build the agents in our AI-native social content operating system, and the deterministic pipelines they run inside , conceiving those workflows as well as writing them.
You'd sit in the core build team, working with the Backend engineers on the services around the pipelines.
What you'll do
- Build and operate multi-step AI pipelines as real workflows , idempotent, retryable, observable, resumable after failure
- Design the evaluation before you ship: datasets, metrics, and gates that can block a release
- Make model behaviour reproducible through versioning, caching and replay, so a decision made months ago can be reconstructed
- Harden pipelines against adversarial and injected input, structurally rather than by prompt instruction
- Keep cost and latency per operation measured, and treat regressions as bugs
- Work with the Site Reliability Engineers on production visibility into provider latency, failure rates and cost
- Keep the model layer swappable, so changing provider is a configuration change rather than a rewrite
What you'll need
- Around 3 or more years in software engineering, including shipping AI or LLM-backed features to production
- Stayed responsible for those features after launch, including their failures
- Built evaluation harnesses and used them to make real ship or no-ship decisions
- Robust general software engineering , most of this job is systems work
- Can reason clearly about the failure modes of a component that doesn't return the same answer twice
Nice to have
- Multi-modal pipelines , vision, audio, or document understanding
- Retrieval systems and the evaluation of retrieval quality
- Adversarial input, prompt injection, or abuse-resistance work
- Fine-tuning or distillation, and the infrastructure around it
- Experience with workflow orchestration at scale.
📌 AI Engineer (Bengaluru)
🏢 PinnacleU
📍 Bengaluru