VK
Hi, I'm Vamsi 👋

VAMSI KRISHNA

_

I build the backend half of AI products — LLM agents that run in parallel, retrieval that cites its sources, and Python services that stay up. 5 years across financial services and autonomous vehicles, shipping production systems, not demos.

View Projects 📄 Résumé in · LinkedIn ⌥ GitHub
01 — About

I build the backend
of AI products

I'm a backend AI engineer. I build the unglamorous half of an AI product: Python/FastAPI services, LLM inference adapters, schema-validated outputs (Pydantic, typed parsers), retrieval that cites its sources, and the eval harnesses that stop a prompt change from quietly breaking production.

At a Fortune 100 Financial-Services Company I built parallel LLM agent workflows for incident intelligence — specialised agents reading operational data concurrently, merged into source-grounded summaries with confidence scoring — which cut complex-case resolution time 40%. On the reliability side I automated multi-region AWS failover for 32 enterprise applications, cutting per-app cutover time 70%, and validated RTO/RPO with Gremlin chaos experiments. Earlier, on autonomous-vehicle programs for a German automotive Tier-1 (ZF), I built ADAS perception models.

Outside work I ship the same ideas as open source — a RAG document assistant with hybrid retrieval and a recall@k eval harness, an LLM token-compression library, a token-aware inference gateway, a chaos-testing toolkit, a Kubernetes failover operator. I like AI that can be measured, and systems that survive a bad day in production.

ROLESoftware Engineer, Backend AI & Reliability @ Fortune 100 Financial Services
FOCUSLLM Agents · RAG · Backend · Reliability
STACKPython · FastAPI · Pydantic · AWS · Docker
EDUM.S. CS — Stevens Institute
EXP5 years
OPEN TOAI / Backend / Platform Engineering
02 — Experience

Where I've shipped

01

Software Engineer · Backend AI Systems & Reliability · Fortune 100 Financial Services

JUL 2023 — PRESENT

Production LLM workflows and multi-region resilience in a regulated financial-services environment.

  • Cut complex-case resolution time 40% with parallel LLM agent workflows — specialised agents analysing operational data concurrently, merged into structured, source-grounded incident summaries with confidence scoring.
  • Cut manual document-processing time 60% with a production generative-AI summarization pipeline delivered across 5 concurrent workstreams.
  • Own the backend for production AI workflows: FastAPI inference adapters, Pydantic schema validation, context assembly, typed response parsers, observability hooks.
  • Cut per-application AWS failover cutover time 70% across 32 enterprise applications with Python runbook automation and pre/post health checks; validated RTO/RPO with Gremlin chaos experiments.
  • Built reusable AI eval and validation harnesses (pytest, schema checks, CI/CD quality gates) — 35% faster integration testing, adopted by 6+ downstream AI projects.
LLM AgentsFastAPIPydanticRAGAWSGremlinCI/CD
02

Machine Learning Engineer · Global IT Services Firm — Client: ZF (Automotive)

JUL 2020 — AUG 2022

Deep-learning perception for autonomous-vehicle ADAS programs.

  • Hit 91% object-detection accuracy translating ADAS requirements into YOLO-based perception models for vehicles, pedestrians and traffic signs in a live inference pipeline.
  • Delivered U-Net and DeepLab semantic-segmentation models in TensorFlow for lane and road-boundary detection across a 5-engineer ML team.
  • Cut segmentation dataset generation time 50% by automating TFRecord serialization and annotation pipelines for camera and LiDAR road-scene data.
  • Surfaced geospatial alignment defects automated tests missed with a Python LiDAR point-cloud visualization tool.
TensorFlowYOLOU-NetDeepLabOpenCVLiDARDocker
03

M.S. Computer Science · Stevens Institute of Technology

2022 — 2024

GPA 3.5 / 4.0 · Hoboken, NJ. Focus on ML systems and software engineering.

03 — What I do

Capabilities

01

AI Agents & LLM Systems

LLM features that ship: agents in parallel, retrieval that cites, outputs that validate.

LLM IntegrationAgentic WorkflowsParallel OrchestrationRAGHybrid RetrievalEmbeddingsVector SearchPrompt EngineeringStructured Output ParsingConfidence ScoringAnthropic / OpenAI APIsHugging Face
02

Backend Engineering

Python services behind the models — typed, tested, observable.

PythonFastAPIFlaskREST APIsPydanticPostgreSQLSQLMicroservicesGoTypeScriptJavaScriptBashGitLinux
03

Evaluation & Quality

Proof that a prompt or model change didn't quietly break anything.

Eval HarnessesGolden Test Setsrecall@kRegression TestingPytestCI Quality GatesPrompt/Model VersioningA/B TestingSchema Validation
04

Cloud & Reliability

Multi-region failover, chaos experiments, reproducible infrastructure.

AWS (EC2/RDS/S3/Lambda/ECS)Route 53CloudWatchTerraformDockerKubernetesGitHub ActionsGremlinDatadogSplunkGrafana
05

ML & Data

Classical models to CV deep nets, trained and evaluated rigorously.

PyTorchTensorFlowScikit-learnLightGBMOpenCVPandasNumPySpark / PySparkKafkaSageMakerU-NetYOLO
04 — Projects

Selected work

01★ FLAGSHIP

Document RAG Assistant

Source-grounded retrieval QA over your own documents

RAG service that answers questions over uploaded documents with citations — or says "I don't know" when the context doesn't cover it. Chunking, embeddings, hybrid keyword + vector retrieval, Claude generation behind FastAPI, and a recall@k eval harness so a retrieval change can be measured, not guessed at. Fully offline-testable with an echo LLM stub.

RAGFastAPIClaude APIHybrid RetrievalEval HarnessDocker
View on GitHub →
02

Incident Intelligence

Paste production logs, get an incident report

Isolation Forest flags the anomalous window, then three LLM agents run in parallel — root cause, blast radius, remediation — and merge into one source-grounded report. The open-source shape of the agent workflow I built at work.

PythonFastAPIIsolation ForestParallel LLM AgentsAIOps
View on GitHub →
03

Inference Gateway

Token-aware routing for LLM traffic

A gateway in front of LLM providers: least-outstanding-tokens balancing (not round-robin), circuit breaking on unhealthy backends, semantic caching, and per-caller cost accounting exported to Prometheus.

PythonFastAPICircuit BreakerSemantic CachePrometheus
View on GitHub →
04

PromptPress

LLM context-compression engine

Zero-dependency Python library that cuts LLM input-token costs 30–70% via a budget-driven compression pipeline — six strategies on an aggressiveness ladder, protected-region shielding, and a drop-in Anthropic SDK middleware.

PythonClaude APIPageRankZero depsCI
View on GitHub →
05

Chaos-Forge

Resilience-testing harness

Injects latency, error and blackhole faults through a reverse proxy, then scores the steady-state hypothesis with p50/p95 probes and a pass/fail resilience report. Stdlib-first, no mocks.

PythonHTTP proxyFault InjectionDockerPytest
View on GitHub →
06

Failover Operator

Kubernetes operator in Go

A FailoverApp CRD that reconciles its Deployment and reports readiness against a minHealthy replica floor — real health, not just "the object exists". Built with kubebuilder / controller-runtime.

GoKubernetesCRDcontroller-runtimekubebuilder
View on GitHub →
07

Credit-Risk MLOps Pipeline

End-to-end reproducible MLOps

Trains, versions and serves a credit-default model behind a schema-validated FastAPI service — seeded runs, no train/serve skew, versioned artifacts, and a CI matrix gating on ROC-AUC thresholds and the full API surface.

Scikit-learnFastAPIPydanticDockerGitHub Actions
View on GitHub →
08

Cloud Deploy IaC

Production-style Terraform on AWS

Ships a containerized service to AWS ECS Fargate behind an ALB with CloudWatch logs, alarms and a dashboard — modular networking/IAM/monitoring, delivered by GitHub Actions with keyless OIDC auth.

TerraformAWS ECSALBOIDCCloudWatch
View on GitHub →
09

Submittal Verifier

Document intelligence for construction specs

Verifies construction submittals against CSI spec sections automatically — flags missing items, standard mismatches and value deviations, turning a manual page-by-page review into a diffable report.

PythonPDF ParsingCSI MasterFormatDocument Automation
View on GitHub →