Vetted Software Engineers in New York

Manasa Pantra

Screened ReferencesStrong rec.

Junior Software Engineer specializing in AI, LLM systems, and full-stack development

Stony Brook, NY2y exp

“Product-focused full-stack engineer at startup (Zippy) who shipped a production multi-agent AI system for restaurant operations plus payments workflows. Built end-to-end: RAG grounded on a Notion knowledge base, structured function-calling task routing, FastAPI/JWT multi-tenant backend, and a polished React+TypeScript owner dashboard. Has real production incident experience (duplicate Stripe webhooks) and reports ~94% task-routing accuracy under load.”

“Internship at ActiveVM where they tackled large-scale Spring Boot 2→3/library migrations across hundreds of downstream products by combining OpenRewrite (AST-based recipes) with an LLM/RAG-based classifier that routed risky files to human experts. Reported ~70% reduction in manual effort and 90%+ accuracy after testing across multiple branches and cutovers; also built a CTR-driven book recommendation capstone showcased at the Google office in Cambridge.”

“Grad student who co-developed a safety-oriented mental health LLM consulting agent using RAG + Gemini and Hugging Face emotion detection to assess user crisis level and adapt responses. Implemented a key reliability improvement for CRISIS scenarios by bypassing generative output and returning direct, emotionless, knowledge-base guidance to seek immediate real-world help.”

“Backend engineer with experience evolving a real-time transaction and rewards processing platform from a tightly coupled architecture into domain-based microservices. Uses REST plus Kafka for synchronous vs. asynchronous workflows, and builds Python/FastAPI APIs with Pydantic contracts, Docker/Kubernetes deployments, and JWT/OAuth-based security; has also supported analytics/dashboard use cases (Power BI).”

“Software engineer with production experience building a serverless monolith and multi-layer video pipeline at easyML, plus hands-on integration of multiple LLM providers (Grok/Claude/OpenAI) into a full-stack app. Interested in robotics via computer vision (OpenCV/OpenMMLab), with a strong real-time systems mindset around SLOs, latency, determinism, and reliability; also has low-level OS experience writing a keyboard device driver.”

“Built and scaled a multi-agent LLM automation pipeline during a fintech internship, growing from a rapid 1-week proof-of-concept to a 15+ agent hierarchical system that cut market brief report generation time from ~5 hours to under 30 minutes. Hands-on with agent frameworks (Haystack, CrewAI, LangChain) and experienced in debugging agent communication issues via sandboxed modular testing and context/token management; also regularly gives architecture-first technical demos at multiple hackathons and university events.”

“Backend engineer with strong performance and security instincts: built a Flask API for readability metrics with clean, testable modular design; optimized SQLAlchemy/Postgres to eliminate N+1 issues (800ms to 120ms). Also implemented an LLM-powered natural-language travel search using Claude Sonnet + Amadeus with RAG and anti-exploitation safeguards, plus multi-tenant isolation via Postgres RLS and Redis caching that cut search latency from ~20s to ~4–5s while reducing storage costs.”

“Solo-built and launched an AI logo generator SaaS in ~2 months using React/Next.js/TypeScript with managed auth and payments, deploying via Vercel/GitHub CI/CD. Also has hands-on AWS production experience running containerized services with Terraform-managed multi-environment infrastructure and strong reliability patterns for integrations/pipelines.”

Daniel Jin

Screened

Intern Site Reliability Engineer specializing in Kubernetes, AWS, and observability

New York, NY1y exp

“Backend/data engineering candidate specializing in Python/Flask services and ML-enabled systems, deploying containerized workloads on AWS ECS/EKS with strong observability (Prometheus/Grafana) and PostgreSQL performance tuning. Built multi-tenant architectures with row- and schema-level isolation and optimized a Kubernetes-based Airflow + Spark nightly ETL pipeline for an e-commerce client, improving performance by 250%+ and reliably beating morning reporting deadlines; also contributed to Apache Airflow (SQLAlchemy/PostgreSQL area).”

“Lead engineer/designer behind The Washington Post's internal live-tracker tooling and election coverage interfaces. They combine cloud architecture, frontend/data-viz craftsmanship, and close newsroom stakeholder collaboration to ship real-time, high-traffic journalism products that improved internal efficiency and supported major audience and subscriber outcomes.”

“Backend/API engineer with ~4 years experience building production services in .NET Core/PostgreSQL/Redis/Docker and optimizing real-world latency issues (claims ~60% response-time improvement). Also built and owned an end-to-end RAG-based AI assistant using Python/FastAPI, OpenAI APIs, and Pinecone, plus agentic workflows with reliability guardrails (retries, confidence thresholds, monitoring). Currently pursuing a master’s degree and targeting a $150k base salary.”

Sanjay Mandru

Screened

Mid-Level Full-Stack Software Engineer specializing in cloud microservices and real-time analytics

Buffalo, NY3y exp

“Software engineer who built a reusable React component package (UI modules, auth helpers, API client wrappers) for an AI SaaS background-removal project, emphasizing performance (tree shaking/dynamic imports) and reliability (Jest + Storybook). Also delivered a unified REST API for Samsung Big Data Portal, resolving cross-team issues by standardizing schemas, improving validation/logging, and operating effectively amid shifting requirements.”

“Built and shipped a production LLM-powered agent that automated purchasing and inventory operations by integrating with live ERP data and returning structured, machine-readable outputs usable by downstream systems. Emphasizes real-world reliability through orchestration, strict schemas/validation, confidence-based fallbacks with human handoff, and monitoring/evaluation feedback loops to reduce silent failures and make issues observable.”