Senior AI Solutions Engineer

Role Overview

• Provision multi-tenant environments: tenant creation, log file type registration, product family configuration, severity thresholds, and API key management.
• Guide customers through LogIQ's Signature Onboarding Wizard.
• Configure per-tenant defaults and document every configuration decision in customer-specific runbooks for long-term maintainability.
• Validate the full detection lifecycle end-to-end on customer log samples before any go-live, including quality benchmarks on hold-out data.
• Set up real-time log stream ingestion pipelines — Kafka, Kinesis, Fluentd, syslog-ng, or customer-native agents — into LogIQ's streaming layer.
• Configure the Anomaly Detection engine: define healthy baselines, tune sensitivity thresholds, and map deviation patterns to specific signature triggers.
• Wire streaming triggers to the RCA Agent so that when an anomaly fires, root-cause investigation begins automatically with no human intervention.
• Monitor stream health: lag, throughput, parsing error rates, and alert on pipeline degradation before it affects customer outcomes.
• Work with customers to identify which log sources to prioritize for streaming vs. batch ingestion, balancing latency requirements against infrastructure cost.
• Ingest and index customer knowledge articles, historical case resolutions, and equipment documentation into the RCA Agent's retrieval layer (OpenSearch + pgvector).
• Configure evidence-weighting rules so the RCA Agent knows which sources to trust most for a given equipment type or failure mode.
• Tune reasoning prompts and retrieval strategies based on observed RCA quality — iterating until root-cause accuracy meets the customer's acceptance criteria.
• Build fix-strategy libraries: map known root causes to recommended remediation steps, pulling from customer SOPs and historical tickets.
• Validate RCA output against historical cases where the true root cause is known; track precision and recall over iteration cycles.
• Ingest, clean, and pre-label customer-provided log samples to build compelling, domain-specific demos that speak directly to the customer's operational pain.
• Demonstrate both reactive (case upload → signature detection → RCA → fix recommendation) and proactive (live stream → anomaly trigger → automated RCA) workflows against real data.
• Create demo scripts, scenario walkthroughs, before/after MTTR comparisons, and leave-behind documentation for prospects.
• Adapt demos quickly to new industries or log types — a customer in manufacturing should see their alarm formats, their fault patterns, their fix vocabulary.
• Design, build, and register new LangGraph agent tools as customer use cases demand — e.g., a tool that queries a customer's CMDB, pulls ticket history from ServiceNow, or fetches firmware changelogs from an internal API.
• Package reusable capabilities as LogIQ Skills: self-contained, versioned bundles of tools, prompts, and configuration that can be applied across customers in the same domain.
• Maintain a tool allowlist and review process so new tools integrate safely with the agent's execution context and tenant isolation guarantees.
• Contribute high-quality tools back to the platform's shared tool library so the whole team benefits.
• Write custom log parsers for proprietary or undocumented equipment formats (Python, plugged into the FastAPI parser registry).
• Build data connectors for customer-specific ingestion sources: REST APIs, SFTP drops, database exports, or cloud storage buckets.
• Define record-splitting rules, type classifiers, and deep-parsed field schemas for new log file types using the Signature Onboarding pipeline.
• Maintain a parser test suite — real sample lines, expected field outputs — so parsers don't regress across platform updates.
• Tune LLM system prompts, memory strategies, context windows, and few-shot examples based on observed agent behavior on customer data.
• Modify the signature workflow DAG to handle customer-specific detection logic that the automated agent generation doesn't cover out of the box.
• Ship targeted bug fixes and feature additions back to the core platform codebase — you are a contributor, not just a consumer.
• Debug async pipeline failures.
• Own the technical relationship for your customer portfolio: onboarding calls, weekly syncs, async Slack/email, and escalation handling.
• Translate customer domain knowledge — telecom alarm semantics, SCADA event codes, IT operations terminology — into LogIQ configuration and agent guidance.
• Train customer teams to operate LogIQ independently: run their own demos, onboard new signatures, and interpret RCA outputs.
• Surface recurring pain points and propose product improvements; your customer exposure gives you signal the core product team cannot get from anywhere else.

Requirements

• 4+ years of production Python. Comfortable with asyncio, FastAPI, Pydantic v2, and SQLAlchemy 2.0. Ability to read and extend an unfamiliar codebase quickly.
• Hands-on experience building or operating LLM-powered agent pipelines — LangChain, LangGraph, CrewAI, AutoGen, or equivalent. Understands state graphs, tool calls, memory, and multi-step reasoning loops.
• Can design, implement, and register new agent tools using the @tool decorator pattern (LangGraph/LangChain). Understands tool allowlists, input/output schemas, and safe integration with existing agent contexts.
• Can systematically diagnose LLM failure modes and improve prompts through controlled iteration. Understands token budgeting, few-shot construction, output format control, and context window management.
• Working knowledge of at least one streaming or log-shipping technology — Kafka, Kinesis, Fluentd, Logstash, syslog-ng, or similar. Understands consumer lag, backpressure, and at-least-once delivery semantics.
• Understands async task queues (Celery, SQS, Redis), message broker patterns, and how to debug distributed pipeline failures from logs and traces.
• Solid PostgreSQL fundamentals: schema design, JSONB queries, indexing. Exposure to time-series stores (TimescaleDB) and full-text search (OpenSearch / Elasticsearch) is a plus.
• Comfortable with AWS (S3, SQS, IAM, Kinesis) or Azure equivalents. Docker and container-based local deployments. Familiarity with docker-compose for multi-service dev environments.
• Strong written and spoken English. Can explain a multi-stage agent failure to a non-technical operations director. Experience in customer-facing technical roles — solutions engineering, implementation, pre-sales, or technical consulting — is a strong plus.
• Education: B.E. / B.Tech or M.Tech in Computer Science, Electronics, or a related engineering discipline. Equivalent industry experience is fully acceptable.

Tech Stack

• AWS
• Azure
• Cloud
• Docker
• ElasticSearch
• Kafka
• Logstash
• Postgres
• Python
• Redis
• ServiceNow