Backend Engineer Intern
Tradl AI
Bengaluru, Karnataka, IndiaFresher1m agovia LinkedIn
This job at a glance
- Who can apply
- Freshers and up to 1 year
- Must know
- NLPWindowsPythonPandasNumPyGit
- Nice to have
- REST APIs
- Work
- On-site in Bengaluru · internship
- Pay
- Not stated by the employer.
- Skill demand
- Python35%NLP3%Pandas2%of 99 similar fresher jobs posted in India this week
- Listed
- 12 Nov 2026 · still open
Sign in free to see which of these skills your resume already shows.
About TRADL AI
TRADL AI is building India's first AI trading co-pilot for 100M+ traders. We convert natural language into verified, code-backed stock intelligence. We are live, growing fast, and focused on solving complex technical challenges.
Why This Role Exists
This internship offers the opportunity to operate across the stack, from fine-tuning models and building agentic workflows to integrating these capabilities into the product. Interns will own real features, push to production, and work directly with the engineering team on challenging problems.
Specialisation Tracks
- Track A: Claude Code & AI Tooling
- Understand agentic workflows.
- Build with skills and sub-agents.
- Configure memory via CLAUDE.md.
- Manage context windows deliberately.
- Wire up MCP servers.
- Use hooks to intercept and extend Claude Code's behaviour.
- Track B: Model Fine-Tuning
- Fine-tuned at least one model end-to-end (data preparation, training, evaluation).
- Understand the difference between full fine-tuning, LoRA, and QLoRA.
- Know when to apply SFT vs alignment methods like DPO.
- Interpret training runs beyond loss reduction.
- Track C: Both (PREFERRED)
- Set up a LoRA fine-tuning run.
- Use Claude Code agents to automate evaluation pipelines.
- Wired hooks, managed context with CLAUDE.md skills, and spun up MCP integrations.
- Understand internal model workings.
Requirements & Tech Stack
- Python proficiency (clean, modular code, scripts, modules, CLI tools).
- Claude Code working knowledge (used for real tasks, not just one-shot prompts).
- Familiarity with CLAUDE.md, skills, and basic tool use.
- Agentic workflow understanding (sub-agents, context window consumption, multi-step task chaining).
- Hands-on experience with Hugging Face (Transformers, datasets, peft, or trl).
- Data wrangling with Pandas and NumPy (building clean instruction datasets).
- Git fluency (branches, PRs, meaningful commit messages).
Bonus Points
- Multi-agent pipelines (orchestrator delegating to sub-agents with context passing and error recovery).
- Hooks (pre/post-tool hooks for logging, filtering, constraint enforcement, or side-effects).
- MCP integration (connecting Claude Code to external tools, databases, or APIs).
- Advanced fine-tuning: QLoRA or LoRA beyond toy examples, preference data + DPO, GRPO, RLVR, or SFT-to-alignment pipeline experiments.