Ankit Maheshwari
Ankit Maheshwari

Ankit Maheshwari

AI/ML • Junior • New Delhi, India • UTC+05:30 • In office, Partially remote, Fully remote • Relocation • Work permits: India • English: C2 • Available in 4 weeks • from $80k per year
DevOps Engineer
9+ years exp
Node JS
Ruby
Bash
JavaScript
Python
SQL
MLOps & Deployment: 5/10
Active 4 days ago
+91 (9632) 130772
Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles

Overview

LLM-focused engineer (senior level) specializing in building agent orchestration and platform-integration backends for LLM-driven workflows. The strongest proven skill is designing agent pipelines and production integrations as shown by the coordinator and agent modules plus FastAPI endpoints and Alembic migrations (opsAgent-AI/backend/app/agents/coordinator.py, opsAgent-AI/backend/app/main.py, opsAgent-AI/backend/alembic/versions/001_initial_schema.py). There is little or no evidence of custom model training, experiment tracking, or advanced efficiency work such as quantization or GPU optimization.

Technical skills

Node JS• Junior
Ruby
Bash
JavaScript
Python• Junior
SQL• Middle • 9y+
Node JS
Express
Python
FastAPI
Alembic
Asyncio
Java
Apache Tomcat• 9y+
Databases
Databases
PostgreSQL• 5y+
AI/ML
Gradio
Google GenAI SDK
OpenAI SDK
Gemini
AI Agents
DevOps
Amazon EC2
Helm
Platform Engineering
Git• 9y+
Nginx• 9y+
Amazon EKS• 5y+
ArgoCD• 5y+
AWS• 5y+
CI/CD• 5y+
Docker• 5y+
Dynatrace• 5y+
GCP• 5y+
GitOps• 5y+
Google GKE• 5y+
Grafana• 5y+
Incident Management• 5y+
Kubernetes• 5y+
Kustomize• 5y+
Prometheus• 5y+
Terraform• 5y+

Timeline

Senior DevOps Engineer • Senior
Jarvis Technology & Consulting • Full-Time
Mar 2021 to Present 5 Years 5 Months New Delhi In office
Owned AWS/GCP cloud infrastructure and a Kubernetes platform supporting 100+ microservices across dev, staging, and production. Implemented GitOps deployments using Argo CD and Kustomize and standardized infrastructure provisioning with reusable Terraform modules. Improved performance by optimizing Kubernetes resource allocation and strengthened security with secure container builds and CI/CD practices. Built observability with Prometheus, Grafana, Dynatrace, and PostgreSQL and led infrastructure migration to separation and modernization goals with minimal downtime.
Kubernetes
Amazon EKS
Google GKE
Docker
Terraform
ArgoCD
Kustomize
Prometheus
Grafana
Dynatrace
PostgreSQL
Git
CI/CD
GitOps
Incident Management
AWS
GCP
Technical Associate • Middle
Amazon India • Full-Time
Jun 2017 to Feb 2021 3 Years 8 Months Bengaluru In office
Configured Nginx and Apache Tomcat on AWS EC2 to support application hosting. Used advanced SQL queries to analyze catalogue data and identify product gaps. Managed GitHub repositories and administered AWS account permissions using IAM roles, users, and access settings.
Nginx
Apache Tomcat
SQL
Gitsince 2017
Narsee Monjee Institute of Management Studies (NMIMS) University
Bachelor's Degree • Electronics
2011–2015 Mumbai, Maharashtra
Junior AI/ML Engineer Confidence: Medium LLM Engineer
LLM-focused engineer (senior level) specializing in building agent orchestration and platform-integration backends for LLM-driven workflows. The strongest proven skill is designing agent pipelines and production integrations as shown by the coordinator and agent modules plus FastAPI endpoints and Alembic migrations (opsAgent-AI/backend/app/agents/coordinator.py, opsAgent-AI/backend/app/main.py, opsAgent-AI/backend/alembic/versions/001_initial_schema.py). There is little or no evidence of custom model training, experiment tracking, or advanced efficiency work such as quantization or GPU optimization.
Model Architecture & Training
2/10
How well models are designed and trained
No custom model architectures or training loops; engineering focuses on prompt design and hosted model orchestration (Gemini/OpenAI) rather than model development.
Evidence
opsAgent-AI/backend/app/agents/coordinator.py: uses agno.Agent and Gemini model integration
multiAgent-tasks/src/agents/base.py: call_openai selects Gemini or OpenAI and builds model requests
opsAgent-AI/backend/app/prompts/coordinator.py: system prompt specifying intent taxonomy and rules
Data Pipeline & Feature Engineering
3/10
How data is prepared for models
Simple data parsing and light scraping utilities exist for email and job posting extraction but no large-scale ETL or feature engineering pipeline.
Evidence
multiAgent-tasks/src/utils/scraper.py: scrape_job using requests and BeautifulSoup
multiAgent-tasks/src/utils/contact_sniffer.py: sniff_contact regex-based contact extraction
opsAgent-AI/backend/app/tasks/email_poller.py: email parsing and extraction logic
Experimentation & Evaluation
2/10
How results are measured and tested
Basic automated tests and unit-style checks are present but there is no experiment tracking, systematic ML evaluation, or reproducible experiment configs.
Evidence
multiAgent-tasks/tests/test_system.py: unit tests for orchestrator, pdf handler and contact sniffer
opsAgent-AI/backend/tests/test_agent.py: tests for intent parsing and agent behavior
MLOps & Deployment
5/10
How models are shipped to production
Production-oriented backend patterns are present: FastAPI endpoints, async SQLAlchemy sessions, Alembic migrations, health/readiness endpoints and dependency injection for DB sessions.
Evidence
opsAgent-AI/backend/app/main.py: FastAPI app creation and lifespan hooks
opsAgent-AI/backend/app/database/session.py: async_sessionmaker and init_db patterns
opsAgent-AI/backend/alembic/versions/001_initial_schema.py: Alembic migrations for schema
Computational Efficiency
1/10
How efficiently computing resources are used
Minimal computational efficiency work; models are called as hosted services and there is no evidence of GPU, quantization, batching, or profiling optimizations.
Evidence
multiAgent-tasks/src/agents/base.py: provider selection logic (Gemini/OpenAI) but no batching or quantization code
opsAgent-AI/backend/app/agents/coordinator.py: _calculate_cost and _extract_token_usage indicate cost awareness but not efficiency engineering
Research Depth & Innovation
1/10
Depth of research and new ideas
No evidence of original research, novel algorithms, or paper reproductions; the work focuses on system integration, prompts and orchestration.
Evidence
opsAgent-AI/backend/app/prompts/coordinator.py: detailed system prompt templates
multiAgent-tasks/src/agents/cover_letter.py: prompt-driven cover letter generation
Expertise
AI Agents & Agentic Workflows• Junior
LLM• Junior
MLOps & Model Lifecycle• Junior
Industries
Artificial Intelligence• Junior
Software• Junior
Technologies
Python• Junior
FastAPI
AI Agents
Google GenAI SDK
Gradio
OpenAI SDK
Gemini
Asyncio
Alembic
Recommendations
  • Develop agent orchestration and platform automation features that parse intents and hand off to deterministic workflows (access approvals, incident triage, GitHub/Gmail automation).
  • Productionize LLM usage monitoring and cost/usage observability (token extraction, cost calculations, SLI/SLO for model calls).
  • Harden approval and safety flows around privileged actions and add end-to-end tests for approval edge cases and failure modes.
  • Add experiment tracking and ML evaluation if building bespoke models, and introduce profiling/efficiency work if on-device or self-hosted inference is planned.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
multiAgent-tasks opsAgent-AI
Junior Backend Developer Confidence: Medium API Engineer
Backend API engineer working at a Middle level with a practical strength in building database-focused admin APIs and tooling. The strongest proven skill is secure-ish PostgreSQL access and role management implemented in the grants and servers routes plus the syncRoles utility (DevOps/pg-access-management-dashboard/backend/src/routes/grants.js and /src/utils/syncRoles.js). There is limited evidence of production-grade observability, large-scale system design, performance testing, or comprehensive security hardening in the human-authored code analyzed.
API Design
3/10
How well APIs are designed
Basic, practical REST API design with input validation and auth middleware but no versioning strategy, pagination patterns, or explicit idempotency keys; error contracts are uneven (some endpoints return structured errors while others rethrow).
Evidence
DevOps/pg-access-management-dashboard/backend/src/routes/grants.js: POST / - express-validator input validation and authenticate middleware
DevOps/pg-access-management-dashboard/backend/src/routes/servers.js: POST /test-connection - explicit connectivity check endpoint and requireAdmin middleware
Data Layer & Database
4/10
Working with databases
Concrete, correct use of transactions, upserts and parameterized queries for many operations and an idempotent role-sync flow, but some SQL is constructed with interpolated identifiers (sanitization attempts present) and there is no explicit migration history in the human-authored files used for scoring.
Evidence
DevOps/pg-access-management-dashboard/backend/src/routes/grants.js: transactional CREATE ROLE/GRANT logic with BEGIN/COMMIT/ROLLBACK
DevOps/pg-access-management-dashboard/backend/src/utils/syncRoles.js: upsert with ON CONFLICT and deletion of stale roles
Scalability & Performance
3/10
Handling load and speed
Some operational concerns addressed (connection pool sizing, connection timeouts, non-blocking role sync) but no systematic caching, queue-based decoupling, load-testing artifacts or rate limiting strategy.
Evidence
DevOps/pg-access-management-dashboard/backend/src/routes/servers.js: testPool config includes max, connectionTimeoutMillis, idleTimeoutMillis
DevOps/pg-access-management-dashboard/backend/src/routes/servers.js: fire-and-forget syncRoles(server.id) after server creation
System Architecture
3/10
Overall system structure
Modular code organization (routes, utils, middleware) and clear separation of concerns for DB interactions and syncing, but it is a small monolith without evidence of deliberate service decomposition or configuration/secret orchestration beyond encrypted password fields.
Evidence
DevOps/pg-access-management-dashboard/backend/src/routes/grants.js: separated grants route handling DB provisioning actions
DevOps/pg-access-management-dashboard/backend/src/utils/syncRoles.js: separate role synchronization utility
Security & Auth
3/10
Protecting data and access
Shows security awareness: auth middleware, input validation, encrypted stored DB passwords and attempts to sanitize role names; however there are risky SQL interpolations and at least one questionable API usage pattern that may allow injection or runtime errors.
Evidence
DevOps/pg-access-management-dashboard/backend/src/routes/grants.js: use of authenticate middleware and express-validator; pg_role sanitized via regex before interpolation
DevOps/pg-access-management-dashboard/backend/src/routes/servers.js: encrypt/decrypt usage for stored DB passwords
Reliability & Observability
3/10
Stability and monitoring
Reasonable reliability primitives are present: transactions with rollback, finally blocks releasing clients and pool.end usage; logging is minimal and structured observability, retry-with-backoff patterns, circuit breakers and comprehensive tests are not present in the analyzed human-authored files.
Evidence
DevOps/pg-access-management-dashboard/backend/src/routes/grants.js: try/catch with client.query('ROLLBACK') and client.release() in finally
DevOps/pg-access-management-dashboard/backend/src/utils/syncRoles.js: try/finally ensuring targetPool.end() and upsert logic
Expertise
Node.js• Junior
Microservices & API Architecture• Junior
Databases & Vector Storage• Junior
Industries
Information Technology• Junior
Technologies
Databases
SQL• Middle • 9y+
Ruby
Node JS• Junior
PostgreSQL• 5y+
Terraform• 5y+
GCP• 5y+
Express
Helm
Dynatrace• 5y+
Kustomize• 5y+
Prometheus• 5y+
CI/CD• 5y+
GitOps• 5y+
ArgoCD• 5y+
Git• 9y+
AWS• 5y+
Docker• 5y+
Kubernetes• 5y+
Nginx• 9y+
Grafana• 5y+
Apache Tomcat• 9y+
Platform Engineering
Amazon EKS• 5y+
Google GKE• 5y+
Amazon EC2
Incident Management• 5y+
Recommendations
  • Develop internal tooling and admin APIs for database access provisioning and lifecycle tasks (role create/revoke, connection tests, sync jobs).
  • Build small-to-medium Node.js backend services that integrate with PostgreSQL and require transactional correctness and role/permission workflows.
  • Implement background workers for scheduled revocation and role synchronization with robust retry, backoff and alerting.
  • Harden input handling and SQL usage: move identifier interpolation to validated whitelists or fully parameterized approaches and add integration tests for edge cases.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
DevOps multiAgent-tasks opsAgent-AI