Resume, Context & Prompts

Configure your background, context, and customize how answers are generated.

Overview

The LLM uses your resume, Q&A Bank, Company/Role Context, and system prompt to generate personalized, relevant answers tailored to your experience and the specific role you're interviewing for.

---

Resume

Your resume provides context for answer generation. The more detail you provide, the better the answers.

How to Add Your Resume

1. Open Settings (gear icon)
2. Find the Resume text area
3. Paste your resume text (not formatted documents)
4. Click the Edit in new window button (external link icon) to open a larger editor if needed

Format Recommendation

Plain text works best. Avoid:

• PDF/Word documents (paste as text instead)
• Excessive formatting
• Very long resumes (aim for1-2 pages of text)

Include:

• Work experience with achievements
• Technical skills
• Education
• Projects
• Certifications

What Gets Included

When generating answers, the LLM receives:

• Your resume
• Company / Role Context (if filled in)
• Q&A Bank (if filled in)
• The system prompt
• Recent transcript (interviewer's questions + your responses)
• Rated examples (thumbs up/down from previous answers)

---

Q&A Bank

Pre-prepared answers and preferences that the LLM references when generating responses. This is useful for questions you already know how you want to answer.

How to Add Q&A Bank Entries

1. Open Settings (gear icon)
2. Find the Q&A Bank text area
3. Enter your pre-prepared answers, key talking points, or preferences
4. Click the Edit in new window button to open a larger editor if needed

Example Q&A Bank

Q: Tell me about yourself
A: I'm a software engineer with 5 years of experience in full-stack development, specializing in React and Node.js.

Q: Why do you want to work here?
A: I'm excited about the company's mission to democratize AI and the opportunity to work on products that scale to millions of users.

Preferences:
- Emphasize leadership experience
- Focus on backend architecture expertise
- Mention open-source contributions when relevant

Tips

• Include both specific Q&As and general preferences
• Add key phrases or terminology you want the LLM to use
• Update before each interview to tailor answers to the specific role

---

Company / Role Context

Company information, job description, and team details that help the LLM tailor answers to the specific role and organization.

How to Add Company Context

1. Open Settings (gear icon)
2. Find the Company / Role Context text area
3. Paste the job description, company info, or team details
4. Click the Edit in new window button to open a larger editor if needed

Example Company Context

Company: TechCorp Inc.
Role: Senior Software Engineer
Team: Platform Team (8 engineers)
Location: Remote (US)

Job Description:
- Design and build microservices architecture
- Mentor junior developers
- Lead technical decisions for the platform team
- Contribute to open-source projects

Tech Stack: Go, Kubernetes, AWS, PostgreSQL, Redis
Company Mission: Building developer tools that make cloud-native development accessible

Tips

• Copy the job description directly from the job posting
• Include the tech stack mentioned in the job listing
• Add company culture notes if available
• Mention the team size and structure if known

---

Edit in New Window

Long text fields in Settings have an Edit in new window button (the external link icon next to the label). This opens a dedicated editor window with more space for comfortable editing.

Supported Fields

Field	Background
Resume	Candidate background and experience
System Prompt	LLM behavior instructions
Q&A Bank	Pre-prepared answers and preferences
Company / Role Context	Role and company details
Extractor System Prompt	Question extraction instructions
Hallucination Filter Phrases	Comma-separated phrases to filter

How It Works

1. Click the button next to any text area label
2. A separate editor window opens with the current text
3. Edit comfortably in the larger window
4. Click Save to apply changes to the main window
5. The editor window closes automatically

Example Resume

John Doe
Software Engineer

Experience:
- Senior Developer at TechCorp (2021-Present)
  - Led development of microservices architecture
  - Improved API response time by 40%
  - Mentored team of 5 developers

- Developer at StartupXYZ (2019-2021)
  - Built real-time data pipeline
  - Implemented CI/CD from scratch

Skills: Python, TypeScript, React, Node.js, PostgreSQL, AWS
Education: BS Computer Science, State University (2019)

---

System Prompt

The system prompt instructs the LLM how to behave when generating answers.

Default Prompt

You are an interview assistant. Answer questions concisely based on the candidate's resume and experience.

Customizing the Prompt

You can modify this to match your interview style. Click the Edit in new window button for more space.

Example: Technical Interview

You are a technical interview assistant. Provide clear, structured answers with:
1. Direct answer to the question
2. Example from the candidate's experience
3. Technical details where relevant
Focus on demonstrating depth of knowledge.

Example: Behavioral Interview

You are a behavioral interview assistant. Use the STAR method:
- Situation: Set the context
- Task: Explain what was needed
- Action: Describe what the candidate did
- Result: Share the outcome
Draw from the resume for specific examples.

Example: Senior/Leadership Role

You are an executive interview assistant. Focus on:
- Strategic thinking
- Leadership examples
- Business impact
- Team growth and mentorship
Provide answers that demonstrate senior-levelthinking.

---

Tips for Better Answers

1. Be Specific in Your Resume

Instead of:

- Built web applications

Use:

- Built React + Node.js web applications serving 100K+ users
- Implemented authentication, real-time notifications, and reporting

2. Include Metrics

Numbers help the LLM generate impactful answers:

• Team sizes managed
• Userbase or revenue impact
• Performance improvements
• Budget responsibility

3. Customize for the Role

Before each interview, update:

• Resume to highlight relevant experience
• Company / Role Context with the job description
• Q&A Bank with pre-prepared answers
• System prompt to match interview type
• Remove outdated or irrelevant information

4. Rate Generated Answers

Use the thumbs up/down feature to help the system learn:

• Good answers are used as future examples
• Bad answers are avoided in future generations

---

Context Window Considerations

The total context sent to the LLM includes:

• System prompt (~50-200 tokens)
• Resume (~500-2000 tokens)
• Company / Role Context (~100-500 tokens)
• Q&A Bank (~200-1000 tokens)
• Rated examples (~100-500 tokens)
• Recent transcript (~500-3000 tokens)
• Generated answer (~200-1000 tokens)

If Context is Too Long

• Shorten your resume
• Condense Company / Role Context to key points
• Reduce Q&A Bank to the most important entries
• Reduce system prompt length
• Use a model with larger context window

---

Answer Rating Memory

Rated answers are stored and used for in-context learning:

• Thumbs up: Added to "good examples" for future prompts
• Thumbs down: Added to "bad examples" to avoid similar responses

Storage Location

Ratings are stored locally:

%APPDATA%/intervu/ratings.json

Clearing Ratings

In Settings → Advanced, you can clear all ratings to start fresh.

---

Next Steps

• Basic Mode — Using Intervu effectively
• Answer Ratings — Learning from your feedback
• LLM Endpoint — Configure answer generation