Yuvaraj Tankala

PostgreSQL

OLTP, migrations, indexing, and careful analytical slices.

MongoDB

Flexible documents and iterative schema for product velocity.

Used at Awair Sense (YC’21): APIs, Data, Cloud

Redis

Caching, rate limits, sessions, and hot-path acceleration.

Kafka

Partitioned logs, stream consumers, and replayable events.

Pulsar

Multi-tenant streaming with geo replication options.

Elasticsearch

Search, observability indices, and text-first retrieval.

OpenSearch

Elastic-compatible stacks with open governance models.

DuckDB

Embedded OLAP for local analytics and fast iteration.

ClickHouse

Columnar analytics and high-cardinality aggregates.

Snowflake

SaaS warehouse, secure sharing, and elastic compute.

BigQuery

Serverless analytics and petabyte-scale SQL in GCP.

dbt

Transformations-as-code and tested warehouse models.

Airflow

DAG orchestration, SLAs, and data pipeline reliability.

Dagster

Software-defined assets and stronger data-aware orchestration.

Apache Spark

Distributed batch and structured streaming at scale.

Apache Flink

Stateful stream processing with event-time correctness.

FastAPI

Async Python APIs, ML microservices, and OpenAPI surfaces.

Django

Batteries-included web apps, admin, and ORM-heavy products.

Flask

Lean WSGI apps and gradual complexity where frameworks should stay small.

Node.js

JavaScript services, tooling, and full-stack glue.

Express

Lightweight HTTP layers and Node microservices.

NestJS

Structured Node backends with DI and modular boundaries.

GraphQL

Schema-first APIs, loaders, and graph efficiency patterns.

tRPC

End-to-end typed RPC popular in modern React/Next stacks.

WebSockets

Realtime feeds, collaborative sessions, and event bridges.

gRPC

Strongly typed RPC between high-throughput internal services.

REST design

Versioning, idempotency, pagination, and explicit error contracts.

Uvicorn

ASGI workers, lifespan hooks, and hot-reload developer loops.

Gunicorn

Process-model WSGI serving for stable Python deployments.

Celery

Distributed queues, retries, and long-running background jobs.

BullMQ

Redis-backed Node job queues with observability.

RabbitMQ

Brokered messaging, routing, and reliable async handoffs.

Temporal

Durable workflows and saga-style orchestration.

Kong

API gateway, plugins, and edge auth rate limits.

NATS

Low-latency pub/sub and cloud-native messaging.

Python

Primary language for ML, LLMs, APIs, research, and production services.

Used at Share Ventures: Screening, Automation, Strategy; CoinDCX: Onboarding, Traffic, Nodes; Awair Sense (YC’21): APIs, Data, Cloud

TypeScript

Typed services, SDKs, and maintainable full-stack boundaries.

JavaScript

Node and browser runtimes for product delivery and tooling.

Go

Concurrent services, cloud CLIs, and latency-sensitive network code.

Rust

Memory-safe components and performance-critical paths.

Java

Enterprise services, JVM ecosystems, and large-team patterns.

C/C++

Algorithms, systems thinking, and low-level performance work.

SQL

Relational modeling, analytics SQL, migrations, and query tuning.

Bash

Automation, packaging, and POSIX deployment scripts.

Kotlin

JVM services, Android-adjacent tooling, and pragmatic typing.

Swift

Apple-platform tooling and performance-aware application code.

Zig

Systems experiments and predictable low-level control.

Ruby

Fast scripting, DSLs, and Rails-era integration stacks.

PHP

Legacy web surfaces and pragmatic request/response workloads.

R

Statistical workflows, research charts, and classical analytics.

Julia

Numerical computing and scientific performance prototyping.

WebAssembly

Portability, edge modules, and browser-adjacent compute.

PyTorch

Training, inference, and research-grade model workflows.

TensorFlow

Graphs, TF Serving, and teams standardized on the TF stack.

JAX

Composable transforms, XLA-minded research, and high-performance ML.

Hugging Face

Transformers, datasets, trainer loops, and hub-driven iteration.

LangChain

Agents, tools, and retrieval-aware orchestration.

LangGraph

Stateful graphs, checkpoints, and multi-step agent control flow.

LlamaIndex

Data connectors, indexing, and retrieval pipelines for LLM apps.

CrewAI

Role-based multi-agent coordination for scoped task teams.

OpenAI API

Chat, embeddings, evals, and assistant-style product APIs.

Anthropic API

Claude-family integrations with safety and tool-use patterns.

vLLM

High-throughput serving, paging, and batched GPU inference.

ONNX

Portable graphs and cross-runtime deployment.

TensorRT

NVIDIA-optimized inference and kernel fusion paths.

Triton Inference Server

Multi-framework serving, dynamic batching, and GPU pools.

scikit-learn

Classical ML, preprocessing, and fast, interpretable baselines.

XGBoost

Gradient boosting for tabular and ranking workloads.

Ray

Distributed training, reinforcement loops, and scalable Python jobs.

MLflow

Experiment tracking, model registry, and lifecycle hooks.

Weights & Biases

Metrics, sweeps, artifacts, and regression visibility.

RAG

Chunking, embeddings, re-ranking, and grounded answers.

Vector DBs

Similarity search at scale and hybrid sparse+dense retrieval.

pgvector

Postgres-native embeddings and SQL-first retrieval.

Pinecone

Managed vector search with low-latency serving SLAs.

Weaviate

Schema-aware vectors, modules, and multi-tenant patterns.

Milvus

Large-scale ANN, sharding, and GPU index options.

Chroma

Lightweight embedding stores for rapid local and prod iteration.

Qdrant

Filtering-rich vector search with strong DevEx and APIs.

Fine-tuning

Domain adaptation, data curation, and compute-efficient training.

LoRA & QLoRA

Parameter-efficient adaptation for LLMs under budget.

DPO & RLHF

Preference optimization and human-feedback alignment loops.

Model evaluation

Benchmarks, LLM-as-judge, and regression gates before ship.

Guardrails

Policy checks, PII filters, and jailbreak-resistant prompting.

MCP servers

Tool surfaces that connect models to vetted enterprise systems.

Agentic workflows

Planning, memory, and safe delegation across tools.

Multimodal models

Vision+audio+text understanding for richer assistants.

Whisper & speech

ASR, diarization, and voice UX for hands-free flows.

Vision transformers

Image encoders, detection adjacency, and CV+LLM fusion.

NumPy

Broadcasting, numerics, and array-first scientific computing.

Pandas

DataFrames, time series, and analyst-friendly transformations.

Polars

Lazy, multithreaded frames for fast local and service ETL.

SciPy

Stats, sparse linear algebra, and classical signal processing.

Matplotlib

Publication-style plots and fine-grained axis control.

Seaborn

Expressive statistical visuals atop the PyData stack.

Plotly

Interactive dashboards and collaborative charting.

Streamlit

Rapid ML demos and stakeholder-facing app shells.

uv

Fast Python package and project management (2025–2026 standard).

Ruff

Unified lint+format speed for large Python repositories.

Black

Opinionated formatting to end bike-shedding in teams.

mypy

Static typing discipline across libraries and services.

Pytest

Fixtures, parametrization, and fast unit plus integration tests.

Poetry

Dependency locking and reproducible library workflows.

Hatch

Modern project tooling for matrices of Python versions.

pre-commit

Hooked checks that keep mainline clean before CI.

DVC

Data versioning beside Git for reproducible ML pipelines.

Cython & Numba

Hot-path acceleration without leaving Python ergonomics.

Jupyter

Exploratory analysis, narrative EDA, and teaching-ready notebooks.

IPython

REPL productivity, magics, and debugger ergonomics.

Typer & Click

Composable CLIs for data jobs and platform tooling.

Rich & Loguru

Structured, beautiful logs and terminal-native UX.

AWS

Compute, IAM, networking, S3-class storage, and managed data adjacency.

Used at Awair Sense (YC’21): APIs, Data, Cloud

GCP

BigQuery, GKE, Vertex-adjacent ML, and multi-cloud patterns.

Used at Awair Sense (YC’21): APIs, Data, Cloud

Azure

Enterprise identity, hybrid cloud, and Microsoft ecosystem hooks.

Docker

Reproducible images, multi-stage builds, and supply-chain hygiene.

Kubernetes

Rollouts, autoscaling, probes, and day-two cluster operations.

Terraform

IaC modules, drift control, and environment parity.

Pulumi

Real-language infra with strong component reuse.

Helm

Charted releases and templated Kubernetes deployments.

GitHub Actions

CI/CD, previews, reusable workflows, and attestations.

GitLab CI

Pipelines, security scans, and self-managed runners.

Argo CD

GitOps rollouts with diff views and progressive delivery.

Linux

Shell fluency, systemd, perf, and production debugging.

Used at Awair Sense (YC’21): APIs, Data, Cloud

Prometheus

Metrics, exposition formats, and Alertmanager patterns.

Grafana

Dashboards, on-call visibility, and data source unification.

Datadog

APM, RUM, synthetics, and unified observability.

Nginx

Reverse proxy, TLS termination, caching, and static offload.

React

Component models, hooks, and interactive product surfaces.

Next.js

SSR/SSG, App Router, edge functions, and SEO-aware apps.

Vue

Progressive adoption and approachable component ergonomics.

Svelte

Compile-time optimizations and small runtime bundles.

Tailwind CSS

Utility-first styling with consistent design tokens.

Vercel

Preview deploys, edge config, and serverless-adjacent hosting.

Playwright

Cross-browser E2E with trace viewers and reliable selectors.

Cypress

Developer-friendly component and integration testing.

Vitest

Fast Vite-native unit tests with modern DX.

Jest

Snapshot and unit coverage across JS/TS codebases.

Storybook

Isolated UI docs, visual testing, and design QA.

Turborepo

Monorepo task caching and scalable build graphs.

Three.js

WebGL scenes, instancing, and immersive 3D UX.

React Three Fiber

Declarative Three.js in React with ecosystem helpers.

Framer Motion

Declarative layout animations and shared-element polish.

ESLint & Prettier

Lint/format gates that keep style and safety consistent.