Back to Notes

Tech Stack Overview

Tech Stack Overview

How to position skills in interviews and on resume. LLM: update when new skills added, certs achieved, or projects shipped.


The Three-Layer Stack

Python (Expert) ─────────────── 5 years, production RAG/LLM systems
Go (Growing) ────────────────── 3 months, goroutines/channels/interfaces
AWS (Production use) ────────── IAM, S3, EC2, Bedrock, Lambda — real usage

Most SDE-3 candidates own one layer. Owning all three = rare. The story: "Python got me here. Go makes me dangerous in backend/infra. AWS makes me deployable end-to-end."


Python — Expert Layer

Depth: 5 years, production systems Domain: LLM/RAG pipelines, backend APIs, data engineering

AreaDepthEvidence
Core languageExpertDecorators, OOP, generators, context managers
AsyncStrongasyncio, aiohttp, event loop internals
LLM/AIStrongLangChain agents, RAG pipelines (AskTGE)
TestingGoodpytest, mocking, integration tests
PerformanceGoodcProfile, memory_profiler, vectorized ops

Interview story: Built and refactored AskTGE — a RAG + multiagent system. Handled SSE streaming, citation handling, performance bottlenecks in a production system. Real complexity, real constraints.

Resources in vault:

  • [[Python/Language Core/Python Programming]]
  • [[Python/Language Core/Object Oriented Programming]]
  • [[Python/Language Core/Decorators]]
  • [[Python/Libraries/Logging]]
  • [[Python/Python Interview Questions]]

Go — Differentiator Layer

Depth: Growing (3–6 months), focused on concurrency and idiomatic patterns Domain: Backend services, CLI tools, API servers

AreaDepthEvidence
Core syntaxSolidVariables, types, structs, interfaces
ConcurrencySolidGoroutines, channels, select, sync.Mutex
HTTPSolidnet/http client, JSON decode, timeouts
InterfacesGoodStructural typing, embedding, polymorphism
GenericsLearningType parameters, constraints

Why Go matters for job switch: Most Python engineers applying to backend roles at Razorpay, Zerodha, CRED, etc. cannot discuss Go at all. Discussing goroutine scheduling vs OS threads in an interview is immediately differentiating.

The angle to use: "Go is my backend language of choice for services where Python's GIL would be a problem — concurrent I/O-heavy services, CLI tools, anything where goroutines make the design cleaner."

Resources in vault:

  • [[Go/Go Topics]] — full index
  • [[Go/Channels]] — goroutines + concurrency
  • [[Go/Interfaces]] — structural typing
  • [[Go/HTTP Clients]] — net/http
  • [[synthesis/Concurrency Deep Dive]] — Go vs Python vs distributed

AWS — Cloud Layer

Depth: Broad foundation — production usage across multiple services Note: MLA-C01 cert deferred until after job switch. AWS is a talking point, not the lead.

ServiceDepthUse case
IAMSolidRoles, policies, assume-role patterns
S3SolidStorage classes, lifecycle, presigned URLs
EC2GoodInstance types, AMI, security groups
VPCGoodSubnets, NAT gateway, security groups
LambdaGoodEvent-driven, cold start, layers
BedrockReal (AskTGE)Managed LLMs, agents, knowledge bases — production use
SageMakerFamiliarTraining, endpoints — not production
Kinesis/GlueFamiliarStreaming, ETL
AthenaFamiliarServerless SQL on S3

Resources in vault:

  • [[AWS/AWS Topics]] — full index

AI/ML — Domain Expertise

This is the moat for targeting AI companies.

AreaDepthEvidence
RAG systemsProductionBuilt AskTGE — ingestion, chunking, vector retrieval, generation
LangChainStrongChains, agents, tools, LCEL → [[AI & ML/Langchain]]
MCPSolidModel Context Protocol, servers, tool calling → [[AI & ML/MCP]]
Vector DBsFamiliarEmbeddings, similarity search, indexing strategies
Prompt engineeringStrongProduction prompts, chain-of-thought, few-shot
AWS BedrockLearningManaged models, agent framework

Resources in vault:

  • [[synthesis/LLM & AI Stack]] — full AI synthesis page
  • [[System Design/Problem Designs/RAG & LLM System]] — system design angle

Resume Positioning Guide

For AI/ML-heavy roles (Sarvam, Krutrim, AI startups)

Lead with: Python + LangChain + RAG production experience + MCP knowledge Mention: AskTGE (private) as "production RAG + multiagent system" De-emphasize: Go (frame as bonus, not primary)

For backend/infra roles (Razorpay, Zerodha, CRED, Groww)

Lead with: Go + Python + System Design coverage Mention: Goroutines/channels knowledge, concurrent system design De-emphasize: ML-heavy content

For cloud/data roles (AWS partner firms, data engineering)

Lead with: Python data stack + Kinesis/Glue/Athena + Bedrock production experience Mention: AWS breadth (IAM→S3→EC2→Lambda→Bedrock), MLA-C01 planned post-offer

For FAANG/Big Tech

Lead with: NeetCode 150 completion, System Design depth, Python + Go Mention: AWS cert, open source (vectorbuilds.dev projects)


Skills NOT on Stack (gaps to know about)

SkillGap levelWhen it matters
Java/KotlinHighFAANG backend, Android
KubernetesMediumPlatform/infra roles
Spark/HadoopLowData engineering heavy roles
TypeScript/Next.jsLowAlready have (portfolio)
RustLowSystems roles (not target)

Related

  • [[synthesis/Job Switch Hub]] — timeline and targets
  • [[synthesis/Interview Prep Hub]] — round-by-round prep
  • [[synthesis/Concurrency Deep Dive]] — Python vs Go deep dive
  • [[synthesis/LLM & AI Stack]] — AI company targeting